[PATCH v1 4/8] Add jsonpath array constructors

agora inbox for [email protected]  
help / color / mirror / Atom feed

[PATCH v1 4/8] Add jsonpath array constructors
265+ messages / 2 participants
[nested] [flat]

* [PATCH v1 4/8] Add jsonpath array constructors
@ 2017-04-06 19:36  Nikita Glukhov <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Nikita Glukhov @ 2017-04-06 19:36 UTC (permalink / raw)

---
 doc/src/sgml/func.sgml                       |  7 ++++
 src/backend/utils/adt/jsonpath.c             | 30 ++++++++++++---
 src/backend/utils/adt/jsonpath_exec.c        | 19 ++++++++++
 src/backend/utils/adt/jsonpath_gram.y        |  2 +
 src/include/utils/jsonpath.h                 |  1 +
 src/test/regress/expected/jsonb_jsonpath.out | 57 ++++++++++++++++++++++++++++
 src/test/regress/expected/jsonpath.out       | 12 ++++++
 src/test/regress/sql/jsonb_jsonpath.sql      | 11 ++++++
 src/test/regress/sql/jsonpath.sql            |  3 ++
 9 files changed, 136 insertions(+), 6 deletions(-)

diff --git a/doc/src/sgml/func.sgml b/doc/src/sgml/func.sgml
index 9defd1a..7347e71 100644
--- a/doc/src/sgml/func.sgml
+++ b/doc/src/sgml/func.sgml
@@ -12632,6 +12632,13 @@ table2-mapping
         <entry><literal>pg $[*], 4, 5</literal></entry>
         <entry><literal>1, 2, 3, 4, 5</literal></entry>
        </row>
+       <row>
+        <entry>Array constructor</entry>
+        <entry>Construct a JSON array by enumeration of its elements enclosed in brackets</entry>
+        <entry><literal>[1, 2, 3]</literal></entry>
+        <entry><literal>pg [$[*], 4, 5]</literal></entry>
+        <entry><literal>[1, 2, 3, 4, 5]</literal></entry>
+       </row>
       </tbody>
      </tgroup>
     </table>
diff --git a/src/backend/utils/adt/jsonpath.c b/src/backend/utils/adt/jsonpath.c
index 0fb4257..16031ba 100644
--- a/src/backend/utils/adt/jsonpath.c
+++ b/src/backend/utils/adt/jsonpath.c
@@ -368,13 +368,19 @@ flattenJsonPathParseItem(JsonPathEncodingContext *cxt, JsonPathParseItem *item,
 		case jpiMinus:
 		case jpiExists:
 		case jpiDatetime:
+		case jpiArray:
 			{
 				int32		arg = reserveSpaceForItemPointer(buf);
 
-				chld = !item->value.arg ? pos :
-					flattenJsonPathParseItem(cxt, item->value.arg,
-											 nestingLevel + argNestingLevel,
-											 insideArraySubscript);
+				if (item->type == jpiArray)
+					checkJsonPathExtensionsEnabled(cxt, item->type);
+
+				if (!item->value.arg)
+					break;
+
+				chld = flattenJsonPathParseItem(cxt, item->value.arg,
+												nestingLevel + argNestingLevel,
+												insideArraySubscript);
 				*(int32 *) (buf->data + arg) = chld - pos;
 			}
 			break;
@@ -780,6 +786,15 @@ printJsonPathItem(StringInfo buf, JsonPathItem *v, bool inKey,
 			if (printBracketes || jspHasNext(v))
 				appendStringInfoChar(buf, ')');
 			break;
+		case jpiArray:
+			appendStringInfoChar(buf, '[');
+			if (v->content.arg)
+			{
+				jspGetArg(v, &elem);
+				printJsonPathItem(buf, &elem, false, false);
+			}
+			appendStringInfoChar(buf, ']');
+			break;
 		default:
 			elog(ERROR, "unrecognized jsonpath item type: %d", v->type);
 	}
@@ -979,6 +994,7 @@ jspInitByBuffer(JsonPathItem *v, char *base, int32 pos)
 		case jpiMinus:
 		case jpiFilter:
 		case jpiDatetime:
+		case jpiArray:
 			read_int32(v->content.arg, base, pos);
 			break;
 		case jpiIndexArray:
@@ -1009,7 +1025,8 @@ jspGetArg(JsonPathItem *v, JsonPathItem *a)
 		   v->type == jpiExists ||
 		   v->type == jpiPlus ||
 		   v->type == jpiMinus ||
-		   v->type == jpiDatetime);
+		   v->type == jpiDatetime ||
+		   v->type == jpiArray);
 
 	jspInitByBuffer(a, v->base, v->content.arg);
 }
@@ -1060,7 +1077,8 @@ jspGetNext(JsonPathItem *v, JsonPathItem *a)
 			   v->type == jpiDatetime ||
 			   v->type == jpiKeyValue ||
 			   v->type == jpiStartsWith ||
-			   v->type == jpiSequence);
+			   v->type == jpiSequence ||
+			   v->type == jpiArray);
 
 		if (a)
 			jspInitByBuffer(a, v->base, v->nextPos);
diff --git a/src/backend/utils/adt/jsonpath_exec.c b/src/backend/utils/adt/jsonpath_exec.c
index 1e31d3c..1dee30a 100644
--- a/src/backend/utils/adt/jsonpath_exec.c
+++ b/src/backend/utils/adt/jsonpath_exec.c
@@ -1175,6 +1175,25 @@ executeItemOptUnwrapTarget(JsonPathExecContext *cxt, JsonPathItem *jsp,
 				break;
 			}
 
+		case jpiArray:
+			{
+				JsonValueList list = {0};
+				JsonbValue *arr;
+
+				if (jsp->content.arg)
+				{
+					jspGetArg(jsp, &elem);
+					res = executeItem(cxt, &elem, jb, &list);
+
+					if (jperIsError(res))
+						break;
+				}
+
+				arr = wrapItemsInArray(&list);
+				res = executeNextItem(cxt, jsp, NULL, arr, found, false);
+				break;
+			}
+
 		default:
 			elog(ERROR, "unrecognized jsonpath item type: %d", jsp->type);
 	}
diff --git a/src/backend/utils/adt/jsonpath_gram.y b/src/backend/utils/adt/jsonpath_gram.y
index 1de6f5a..3507c1f 100644
--- a/src/backend/utils/adt/jsonpath_gram.y
+++ b/src/backend/utils/adt/jsonpath_gram.y
@@ -217,6 +217,8 @@ path_primary:
 	| '@'							{ $$ = makeItemType(jpiCurrent); }
 	| LAST_P						{ $$ = makeItemType(jpiLast); }
 	| '(' expr_seq ')'				{ $$ = $2; }
+	| '[' ']'						{ $$ = makeItemUnary(jpiArray, NULL); }
+	| '[' expr_or_seq ']'			{ $$ = makeItemUnary(jpiArray, $2); }
 	;
 
 accessor_expr:
diff --git a/src/include/utils/jsonpath.h b/src/include/utils/jsonpath.h
index 4cbdce3..1b2b964 100644
--- a/src/include/utils/jsonpath.h
+++ b/src/include/utils/jsonpath.h
@@ -88,6 +88,7 @@ typedef enum JsonPathItemType
 	jpiStartsWith,				/* STARTS WITH predicate */
 	jpiLikeRegex,				/* LIKE_REGEX predicate */
 	jpiSequence,				/* sequence constructor: 'expr, ...' */
+	jpiArray,					/* array constructor: '[expr, ...]' */
 } JsonPathItemType;
 
 /* XQuery regex mode flags for LIKE_REGEX predicate */
diff --git a/src/test/regress/expected/jsonb_jsonpath.out b/src/test/regress/expected/jsonb_jsonpath.out
index c5252ad..cd56876 100644
--- a/src/test/regress/expected/jsonb_jsonpath.out
+++ b/src/test/regress/expected/jsonb_jsonpath.out
@@ -2199,6 +2199,12 @@ SELECT jsonb_path_query('[{"a": 1}, {"a": 2}]', '$[*] ? (@.a > 10)');
 ------------------
 (0 rows)
 
+SELECT jsonb_path_query('[{"a": 1}, {"a": 2}]', 'pg [$[*].a]');
+ jsonb_path_query 
+------------------
+ [1, 2]
+(1 row)
+
 SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a');
 ERROR:  JSON object does not contain key "a"
 SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a');
@@ -2231,6 +2237,12 @@ SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].
  []
 (1 row)
 
+SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}]', 'pg [$[*].a]');
+ jsonb_path_query_array 
+------------------------
+ [[1, 2]]
+(1 row)
+
 SELECT jsonb_path_query_first('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a');
 ERROR:  JSON object does not contain key "a"
 SELECT jsonb_path_query_first('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a', silent => true);
@@ -2578,3 +2590,48 @@ select jsonb_path_query('[1,2,3,4,5]', 'pg $[(0, $[*], 5) ? (@ == 3)]');
 
 select jsonb_path_query('[1,2,3,4,5]', 'pg $[(0, $[*], 3) ? (@ == 3)]');
 ERROR:  jsonpath array subscript is not a single numeric value
+-- extension: array constructors
+select jsonb_path_query('[1, 2, 3]', 'pg []');
+ jsonb_path_query 
+------------------
+ []
+(1 row)
+
+select jsonb_path_query('[1, 2, 3]', 'pg [1, 2, $[*], 4, 5]');
+   jsonb_path_query    
+-----------------------
+ [1, 2, 1, 2, 3, 4, 5]
+(1 row)
+
+select jsonb_path_query('[1, 2, 3]', 'pg [1, 2, $[*], 4, 5][*]');
+ jsonb_path_query 
+------------------
+ 1
+ 2
+ 1
+ 2
+ 3
+ 4
+ 5
+(7 rows)
+
+select jsonb_path_query('[1, 2, 3]', 'pg [(1, (2, $[*])), (4, 5)]');
+   jsonb_path_query    
+-----------------------
+ [1, 2, 1, 2, 3, 4, 5]
+(1 row)
+
+select jsonb_path_query('[1, 2, 3]', 'pg [[1, 2], [$[*], 4], 5, [(1,2)?(@ > 5)]]');
+       jsonb_path_query        
+-------------------------------
+ [[1, 2], [1, 2, 3, 4], 5, []]
+(1 row)
+
+select jsonb_path_query('[1, 2, 3]', 'pg strict [1, 2, $[*].a, 4, 5]');
+ERROR:  jsonpath member accessor can only be applied to an object
+select jsonb_path_query('[[1, 2], [3, 4, 5], [], [6, 7]]', 'pg [$[*][*] ? (@ > 3)]');
+ jsonb_path_query 
+------------------
+ [4, 5, 6, 7]
+(1 row)
+
diff --git a/src/test/regress/expected/jsonpath.out b/src/test/regress/expected/jsonpath.out
index f6a132c..e212deb 100644
--- a/src/test/regress/expected/jsonpath.out
+++ b/src/test/regress/expected/jsonpath.out
@@ -604,6 +604,18 @@ select 'pg $[(1, (2, $.a)), 3, (4, 5)]'::jsonpath;
  pg $[(1, (2, $."a")),3,(4, 5)]
 (1 row)
 
+select 'pg []'::jsonpath;
+ jsonpath 
+----------
+ pg []
+(1 row)
+
+select 'pg [[1, 2], ([(3, 4, 5), 6], []), $.a[*]]'::jsonpath;
+                  jsonpath                   
+---------------------------------------------
+ pg [[1, 2], ([(3, 4, 5), 6], []), $."a"[*]]
+(1 row)
+
 select '$ ? (@.a < 1)'::jsonpath;
    jsonpath    
 ---------------
diff --git a/src/test/regress/sql/jsonb_jsonpath.sql b/src/test/regress/sql/jsonb_jsonpath.sql
index 0619ade..be5c7c8 100644
--- a/src/test/regress/sql/jsonb_jsonpath.sql
+++ b/src/test/regress/sql/jsonb_jsonpath.sql
@@ -525,6 +525,7 @@ set time zone default;
 
 SELECT jsonb_path_query('[{"a": 1}, {"a": 2}]', '$[*]');
 SELECT jsonb_path_query('[{"a": 1}, {"a": 2}]', '$[*] ? (@.a > 10)');
+SELECT jsonb_path_query('[{"a": 1}, {"a": 2}]', 'pg [$[*].a]');
 
 SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a');
 SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a');
@@ -532,6 +533,7 @@ SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ == 1)');
 SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}]', '$[*].a ? (@ > 10)');
 SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', vars => '{"min": 1, "max": 4}');
 SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}, {"a": 3}, {"a": 5}]', '$[*].a ? (@ > $min && @ < $max)', vars => '{"min": 3, "max": 4}');
+SELECT jsonb_path_query_array('[{"a": 1}, {"a": 2}]', 'pg [$[*].a]');
 
 SELECT jsonb_path_query_first('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a');
 SELECT jsonb_path_query_first('[{"a": 1}, {"a": 2}, {}]', 'strict $[*].a', silent => true);
@@ -587,3 +589,12 @@ select jsonb_path_query('[1,2,3,4,5]', 'pg -(10, 20, $[1 to 3], 30)');
 select jsonb_path_query('[1,2,3,4,5]', 'pg lax (10, 20.5, $[1 to 3], "30").double()');
 select jsonb_path_query('[1,2,3,4,5]', 'pg $[(0, $[*], 5) ? (@ == 3)]');
 select jsonb_path_query('[1,2,3,4,5]', 'pg $[(0, $[*], 3) ? (@ == 3)]');
+
+-- extension: array constructors
+select jsonb_path_query('[1, 2, 3]', 'pg []');
+select jsonb_path_query('[1, 2, 3]', 'pg [1, 2, $[*], 4, 5]');
+select jsonb_path_query('[1, 2, 3]', 'pg [1, 2, $[*], 4, 5][*]');
+select jsonb_path_query('[1, 2, 3]', 'pg [(1, (2, $[*])), (4, 5)]');
+select jsonb_path_query('[1, 2, 3]', 'pg [[1, 2], [$[*], 4], 5, [(1,2)?(@ > 5)]]');
+select jsonb_path_query('[1, 2, 3]', 'pg strict [1, 2, $[*].a, 4, 5]');
+select jsonb_path_query('[[1, 2], [3, 4, 5], [], [6, 7]]', 'pg [$[*][*] ? (@ > 3)]');
diff --git a/src/test/regress/sql/jsonpath.sql b/src/test/regress/sql/jsonpath.sql
index 2a53168..23ae376 100644
--- a/src/test/regress/sql/jsonpath.sql
+++ b/src/test/regress/sql/jsonpath.sql
@@ -113,6 +113,9 @@ select 'pg (1, 2, $.a) == 5'::jsonpath;
 select 'pg $[(1, 2, $.a) to (3, 4)]'::jsonpath;
 select 'pg $[(1, (2, $.a)), 3, (4, 5)]'::jsonpath;
 
+select 'pg []'::jsonpath;
+select 'pg [[1, 2], ([(3, 4, 5), 6], []), $.a[*]]'::jsonpath;
+
 select '$ ? (@.a < 1)'::jsonpath;
 select '$ ? (@.a < -1)'::jsonpath;
 select '$ ? (@.a < +1)'::jsonpath;
-- 
2.7.4


--------------574DCE16CAD28E71E2083DCD
Content-Type: text/x-patch;
 name="v1-0005-Add-jsonpath-object-constructors.patch"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
 filename="v1-0005-Add-jsonpath-object-constructors.patch"



^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

* [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source
@ 2025-07-27 15:48  Lukas Fittl <[email protected]>
  0 siblings, 0 replies; 265+ messages in thread

From: Lukas Fittl @ 2025-07-27 15:48 UTC (permalink / raw)

In passing also reduce the per-loop overhead caused by repeated divisions
in INSTR_TIME_GET_NANOSEC when the ticks variable has become very large,
instead diff first and then turn it into nanosecs.
---
 src/bin/pg_test_timing/pg_test_timing.c | 59 +++++++++++++++++++------
 src/include/portability/instr_time.h    | 30 ++++++++-----
 2 files changed, 65 insertions(+), 24 deletions(-)

diff --git a/src/bin/pg_test_timing/pg_test_timing.c b/src/bin/pg_test_timing/pg_test_timing.c
index a5621251afc..b77ef2063b6 100644
--- a/src/bin/pg_test_timing/pg_test_timing.c
+++ b/src/bin/pg_test_timing/pg_test_timing.c
@@ -16,6 +16,7 @@ static const char *progname;
 
 static unsigned int test_duration = 3;
 static double max_rprct = 99.99;
+static bool fast_timing = false;
 
 /* record duration in powers of 2 nanoseconds */
 static long long int histogram[32];
@@ -56,6 +57,7 @@ handle_args(int argc, char *argv[])
 	static struct option long_options[] = {
 		{"duration", required_argument, NULL, 'd'},
 		{"cutoff", required_argument, NULL, 'c'},
+		{"fast", no_argument, NULL, 'f'},
 		{NULL, 0, NULL, 0}
 	};
 
@@ -68,7 +70,7 @@ handle_args(int argc, char *argv[])
 	{
 		if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-?") == 0)
 		{
-			printf(_("Usage: %s [-d DURATION] [-c CUTOFF]\n"), progname);
+			printf(_("Usage: %s [-d DURATION] [-c CUTOFF] [--fast]\n"), progname);
 			exit(0);
 		}
 		if (strcmp(argv[1], "--version") == 0 || strcmp(argv[1], "-V") == 0)
@@ -78,7 +80,7 @@ handle_args(int argc, char *argv[])
 		}
 	}
 
-	while ((option = getopt_long(argc, argv, "d:c:",
+	while ((option = getopt_long(argc, argv, "d:c:f:",
 								 long_options, &optindex)) != -1)
 	{
 		switch (option)
@@ -125,6 +127,10 @@ handle_args(int argc, char *argv[])
 				}
 				break;
 
+			case 'f':
+				fast_timing = true;
+				break;
+
 			default:
 				fprintf(stderr, _("Try \"%s --help\" for more information.\n"),
 						progname);
@@ -155,11 +161,31 @@ test_timing(unsigned int duration)
 	uint64		total_time;
 	int64		time_elapsed = 0;
 	uint64		loop_count = 0;
-	uint64		prev,
-				cur;
 	instr_time	start_time,
 				end_time,
-				temp;
+				prev,
+				cur;
+	char	   *time_source = NULL;
+	bool		fast_timing_used = false;
+
+	INSTR_TIME_INITIALIZE();
+
+#if !defined(WIN32) && defined(__x86_64__) && defined(__linux__)
+	if (fast_timing && has_rdtsc)
+	{
+		time_source = "RDTSC";
+		fast_timing_used = true;
+	}
+	else if (has_rdtscp)
+		time_source = "RDTSCP";
+	else
+		time_source = PG_INSTR_CLOCK_NAME;
+#else
+	time_source = PG_INSTR_CLOCK_NAME;
+#endif
+	if (fast_timing && !fast_timing_used)
+		printf(_("Warning: Fast timing requested, but not available - regular timing source will be used\n"));
+	printf(_("Time source: %s\n"), time_source);
 
 	/*
 	 * Pre-zero the statistics data structures.  They're already zero by
@@ -173,8 +199,11 @@ test_timing(unsigned int duration)
 
 	total_time = duration > 0 ? duration * INT64CONST(1000000000) : 0;
 
-	INSTR_TIME_SET_CURRENT(start_time);
-	cur = INSTR_TIME_GET_NANOSEC(start_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(start_time);
+	else
+		INSTR_TIME_SET_CURRENT(start_time);
+	cur = start_time;
 
 	while (time_elapsed < total_time)
 	{
@@ -182,9 +211,11 @@ test_timing(unsigned int duration)
 					bits;
 
 		prev = cur;
-		INSTR_TIME_SET_CURRENT(temp);
-		cur = INSTR_TIME_GET_NANOSEC(temp);
-		diff = cur - prev;
+		if (fast_timing)
+			INSTR_TIME_SET_CURRENT_FAST(cur);
+		else
+			INSTR_TIME_SET_CURRENT(cur);
+		diff = INSTR_TIME_DIFF_NANOSEC(cur, prev);
 
 		/* Did time go backwards? */
 		if (unlikely(diff < 0))
@@ -217,11 +248,13 @@ test_timing(unsigned int duration)
 			largest_diff_count++;
 
 		loop_count++;
-		INSTR_TIME_SUBTRACT(temp, start_time);
-		time_elapsed = INSTR_TIME_GET_NANOSEC(temp);
+		time_elapsed = INSTR_TIME_DIFF_NANOSEC(cur, start_time);
 	}
 
-	INSTR_TIME_SET_CURRENT(end_time);
+	if (fast_timing)
+		INSTR_TIME_SET_CURRENT_FAST(end_time);
+	else
+		INSTR_TIME_SET_CURRENT(end_time);
 
 	INSTR_TIME_SUBTRACT(end_time, start_time);
 
diff --git a/src/include/portability/instr_time.h b/src/include/portability/instr_time.h
index e2e339a0c4f..f02296f1026 100644
--- a/src/include/portability/instr_time.h
+++ b/src/include/portability/instr_time.h
@@ -112,10 +112,13 @@ extern int64 max_ticks_no_overflow;
  */
 #if defined(__darwin__) && defined(CLOCK_MONOTONIC_RAW)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC_RAW
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC_RAW)"
 #elif defined(CLOCK_MONOTONIC)
 #define PG_INSTR_CLOCK	CLOCK_MONOTONIC
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_MONOTONIC)"
 #else
 #define PG_INSTR_CLOCK	CLOCK_REALTIME
+#define PG_INSTR_CLOCK_NAME	"clock_gettime (CLOCK_REALTIME)"
 #endif
 
 #if defined(__x86_64__) && defined(__linux__)
@@ -174,7 +177,7 @@ pg_get_ticks(void)
 }
 
 static inline int64_t
-pg_ticks_to_ns(instr_time t)
+pg_ticks_to_ns(int64 ticks)
 {
 	/*
 	 * Would multiplication overflow? If so perform computation in two parts.
@@ -183,7 +186,7 @@ pg_ticks_to_ns(instr_time t)
 	 */
 	int64		ns = 0;
 
-	if (unlikely(t.ticks > max_ticks_no_overflow))
+	if (unlikely(ticks > max_ticks_no_overflow))
 	{
 		/*
 		 * Compute how often the maximum number of ticks fits completely into
@@ -192,7 +195,7 @@ pg_ticks_to_ns(instr_time t)
 		 * value. In a 2nd step we adjust the number of elapsed ticks and
 		 * convert the remaining ticks.
 		 */
-		int64		count = t.ticks / max_ticks_no_overflow;
+		int64		count = ticks / max_ticks_no_overflow;
 		int64		max_ns = max_ticks_no_overflow * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 
 		ns = max_ns * count;
@@ -201,11 +204,11 @@ pg_ticks_to_ns(instr_time t)
 		 * Subtract the ticks that we now already accounted for, so that they
 		 * don't get counted twice.
 		 */
-		t.ticks -= count * max_ticks_no_overflow;
-		Assert(t.ticks >= 0);
+		ticks -= count * max_ticks_no_overflow;
+		Assert(ticks >= 0);
 	}
 
-	ns += t.ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
+	ns += ticks * ticks_per_ns_scaled / TICKS_TO_NS_PRECISION;
 	return ns;
 }
 
@@ -226,14 +229,14 @@ pg_initialize_get_ticks()
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_get_ticks())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	pg_ticks_to_ns(t)
-
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	(pg_ticks_to_ns(ticks))
 
 #else							/* WIN32 */
 
 
 /* Use QueryPerformanceCounter() */
+#define PG_INSTR_CLOCK_NAME	"QueryPerformanceCounter"
 
 /* helper for INSTR_TIME_SET_CURRENT / INSTR_TIME_SET_CURRENT_FAST */
 static inline instr_time
@@ -265,8 +268,8 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SET_CURRENT(t) \
 	((t) = pg_query_performance_counter())
 
-#define INSTR_TIME_GET_NANOSEC(t) \
-	((int64) ((t).ticks * ((double) NS_PER_S / GetTimerFrequency())))
+#define INSTR_TIME_TICKS_TO_NANOSEC(ticks) \
+	((int64) ((ticks) * ((double) NS_PER_S / GetTimerFrequency())))
 
 #endif							/* WIN32 */
 
@@ -285,9 +288,14 @@ GetTimerFrequency(void)
 #define INSTR_TIME_SUBTRACT(x,y) \
 	((x).ticks -= (y).ticks)
 
+#define INSTR_TIME_DIFF_NANOSEC(x,y) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((x).ticks - (y).ticks))
+
 #define INSTR_TIME_ACCUM_DIFF(x,y,z) \
 	((x).ticks += (y).ticks - (z).ticks)
 
+#define INSTR_TIME_GET_NANOSEC(t) \
+	(INSTR_TIME_TICKS_TO_NANOSEC((t).ticks))
 
 #define INSTR_TIME_GET_DOUBLE(t) \
 	((double) INSTR_TIME_GET_NANOSEC(t) / NS_PER_S)
-- 
2.47.3


--vtqqtrpooseurzip--





^ permalink  raw  reply  [nested|flat] 265+ messages in thread

end of thread, other threads:[~2025-07-27 15:48 UTC | newest]

Thread overview: 265+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2017-04-06 19:36 [PATCH v1 4/8] Add jsonpath array constructors Nikita Glukhov <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>
2025-07-27 15:48 [PATCH v12 3/3] pg_test_timing: Add --fast flag to test fast timing, report time source Lukas Fittl <[email protected]>

This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox