MIME-Version: 1.0
References: <cf85f46f-b02f-05b2-5248-5000b894ebab@enterprisedb.com>
 <20230609000600.syqy447e6metnvyj@awork3.anarazel.de> <a062ac09-c574-1d6a-22b6-f3cc0d28ed42@enterprisedb.com>
 <20230610203456.5gancfekm4pj4pbs@awork3.anarazel.de> <6030d836-e8b7-e7b9-2cbb-144309679d03@enterprisedb.com>
 <8c86c3a6-074e-6c88-3e7e-9452b6a37b9b@enterprisedb.com> <3cd40425-965a-5ce1-1af3-d51971c44b93@enterprisedb.com>
 <8ec36f51-b863-60e3-20e2-b9c981c5ce5e@enterprisedb.com> <06bb7d02-2c44-3062-731e-a735ba13da7e@enterprisedb.com>
 <367160ea-b1ed-4481-e804-bca509128878@enterprisedb.com> <cd63c8bc-3d3c-00ab-f7eb-fa55969c3234@enterprisedb.com>
 <CA+Tgmoaf0EHzcfWRydntzqhfc1Nv96C-60fEBbdf1pYZmT0SLA@mail.gmail.com> <280dc83c-a16f-4424-1319-95e7e3f798bd@enterprisedb.com>
In-Reply-To: <280dc83c-a16f-4424-1319-95e7e3f798bd@enterprisedb.com>
From: Dilip Kumar <dilipbalaut@gmail.com>
Date: Thu, 21 Dec 2023 12:19:20 +0530
Message-ID: <CAFiTN-v1DYMwibRHPnyBtkkQ8-S53G=QXMmpz3sVYgU0w9tsaQ@mail.gmail.com>
Subject: Re: index prefetching
To: Tomas Vondra <tomas.vondra@enterprisedb.com>
Cc: Robert Haas <robertmhaas@gmail.com>, Andres Freund <andres@anarazel.de>, 
	PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Georgios <gkokolatos@protonmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: <https://www.postgresql.org/message-id/CAFiTN-v1DYMwibRHPnyBtkkQ8-S53G%3DQXMmpz3sVYgU0w9tsaQ%40mail.gmail.com>
Precedence: bulk

On Wed, Dec 20, 2023 at 7:11=E2=80=AFAM Tomas Vondra
<tomas.vondra@enterprisedb.com> wrote:
>
I was going through to understand the idea, couple of observations

--
+ for (int i =3D 0; i < PREFETCH_LRU_SIZE; i++)
+ {
+ entry =3D &prefetch->prefetchCache[lru * PREFETCH_LRU_SIZE + i];
+
+ /* Is this the oldest prefetch request in this LRU? */
+ if (entry->request < oldestRequest)
+ {
+ oldestRequest =3D entry->request;
+ oldestIndex =3D i;
+ }
+
+ /*
+ * If the entry is unused (identified by request being set to 0),
+ * we're done. Notice the field is uint64, so empty entry is
+ * guaranteed to be the oldest one.
+ */
+ if (entry->request =3D=3D 0)
+ continue;

If the 'entry->request =3D=3D 0' then we should break instead of continue, =
right?

---
/*
 * Used to detect sequential patterns (and disable prefetching).
 */
#define PREFETCH_QUEUE_HISTORY 8
#define PREFETCH_SEQ_PATTERN_BLOCKS 4

If for sequential patterns we search only 4 blocks then why we are
maintaining history for 8 blocks

---

+ *
+ * XXX Perhaps this should be tied to effective_io_concurrency somehow?
+ *
+ * XXX Could it be harmful that we read the queue backwards? Maybe memory
+ * prefetching works better for the forward direction?
+ */
+ for (int i =3D 1; i < PREFETCH_SEQ_PATTERN_BLOCKS; i++)

Correct, I think if we fetch this forward it will have an advantage
with memory prefetching.

--=20
Regards,
Dilip Kumar
EnterpriseDB: http://www.enterprisedb.com