From: Ronan Dunklau <ronan.dunklau@aiven.io>
To: David Rowley <dgrowleyml@gmail.com>
To: John Naylor <john.naylor@enterprisedb.com>
Cc: James Coleman <jtc331@gmail.com>
Cc: PostgreSQL Developers <pgsql-hackers@lists.postgresql.org>
Cc: Ranier Vilela <ranier.vf@gmail.com>
Cc: Dilip Kumar <dilipbalaut@gmail.com>
Subject: Re: [PATCH] Use optimized single-datum tuplesort in ExecSort
Date: Thu, 15 Jul 2021 12:18:22 +0200
Message-ID: <2502562.vDctQ7PNa3@aivenronan> (raw)
In-Reply-To: <CAFBsxsEegvJsvUUL-rr7=-gE92=ku5w7FQHK6694VRNFt=SSwQ@mail.gmail.com>
References: <3177670.itZtoPt7T5@aivenronan>
<CAApHDvpeQLbiXQYrr+OQAAAyOSgqdroDo6aYAEU1h3XtDSyABA@mail.gmail.com>
<CAFBsxsEegvJsvUUL-rr7=-gE92=ku5w7FQHK6694VRNFt=SSwQ@mail.gmail.com>
Le jeudi 15 juillet 2021, 01:30:26 CEST John Naylor a écrit :
> On Wed, Jul 14, 2021 at 6:14 AM David Rowley <dgrowleyml@gmail.com> wrote:
> > It would be good to get a 2nd opinion about this idea. Also, more
> > benchmark results with v6 and v8 would be good too.
>
Hello,
Thank you for trying this approach in v8 David !
I've decided to test on more "stable" hardware, an EC-2 medium instance,
compiling with Debian's gcc 8.3. That's still not ideal but a lot better than
a laptop.
To gather more meaningful results, I ran every pgbench for 30s instead of the
10 in the initial script provided by David. I ran the full script once for
HEAD, v6, v8, then a second time for HEAD, v6, v8 to try to eliminate noise
that could happen for 90 consecutive seconds, and took for each of those the
median of the 6 runs. It's much less noisy than my previous runs but still
not as as stable as I'd like to.
The results are attached in graph form, as well as the raw data if someone
wants it.
As a conclusion, I don't think it's worth it to introduce a separate
execprocnode function for that case. It is likely the minor difference still
observed can be explained to noise, as they fluctuate if you compare the min,
max, average or median values from the results.
Best regards,
--
Ronan Dunklau
Attachments:results.csv (text/csv, 2.7K) download
name,querynumber,run,value
master,1,1,448.134618
master,1,2,449.589985
master,1,3,450.619392
master,1,4,451.835713
master,1,5,453.799955
master,1,6,449.656700
master,2,1,160.304157
master,2,2,159.194826
master,2,3,160.486337
master,2,4,157.327040
master,2,5,149.773621
master,2,6,156.001160
master,3,1,300.677621
master,3,2,300.906490
master,3,3,300.948030
master,3,4,300.494371
master,3,5,300.236476
master,3,6,300.537330
master,4,1,131.364657
master,4,2,120.390513
master,4,3,125.846061
master,4,4,131.216385
master,4,5,131.036531
master,4,6,132.669908
master,5,1,202.268035
master,5,2,203.257486
master,5,3,202.559354
master,5,4,202.392410
master,5,5,201.532070
master,5,6,196.208607
master,6,1,178.584481
master,6,2,177.955467
master,6,3,178.801931
master,6,4,174.769964
master,6,5,178.523056
master,6,6,178.089879
master,7,1,113.231133
master,7,2,113.978583
master,7,3,107.244092
master,7,4,113.309029
master,7,5,114.732538
master,7,6,112.608871
master,8,1,374.876726
master,8,2,377.042935
master,8,3,377.417160
master,8,4,379.001567
master,8,5,379.139665
master,8,6,378.489882
v6,1,1,774.599303
v6,1,2,794.282738
v6,1,3,794.573274
v6,1,4,786.462616
v6,1,5,793.436419
v6,1,6,794.077087
v6,2,1,181.684654
v6,2,2,181.511128
v6,2,3,171.686299
v6,2,4,156.607679
v6,2,5,150.534859
v6,2,6,153.857928
v6,3,1,437.601327
v6,3,2,438.129653
v6,3,3,438.291123
v6,3,4,440.916755
v6,3,5,441.174564
v6,3,6,440.738299
v6,4,1,147.996631
v6,4,2,148.126633
v6,4,3,146.419228
v6,4,4,130.344895
v6,4,5,131.744894
v6,4,6,133.263237
v6,5,1,225.757004
v6,5,2,225.600496
v6,5,3,225.576930
v6,5,4,226.347284
v6,5,5,224.425254
v6,5,6,218.768964
v6,6,1,196.128685
v6,6,2,197.337412
v6,6,3,198.729160
v6,6,4,197.664255
v6,6,5,200.419565
v6,6,6,200.076977
v6,7,1,125.414408
v6,7,2,122.788568
v6,7,3,124.698258
v6,7,4,112.768749
v6,7,5,113.066527
v6,7,6,111.011427
v6,8,1,643.292463
v6,8,2,644.358034
v6,8,3,643.875166
v6,8,4,642.763715
v6,8,5,642.806728
v6,8,6,641.536225
v8,1,1,780.693054
v8,1,2,780.707289
v8,1,3,779.381265
v8,1,4,755.864425
v8,1,5,790.653100
v8,1,6,788.112981
v8,2,1,159.354465
v8,2,2,157.702581
v8,2,3,159.168693
v8,2,4,154.088551
v8,2,5,158.638991
v8,2,6,160.611918
v8,3,1,442.174582
v8,3,2,442.874461
v8,3,3,443.113160
v8,3,4,443.669698
v8,3,5,444.011233
v8,3,6,443.424492
v8,4,1,133.106784
v8,4,2,130.355708
v8,4,3,132.366164
v8,4,4,131.248541
v8,4,5,131.318346
v8,4,6,131.119015
v8,5,1,225.082939
v8,5,2,225.890904
v8,5,3,225.603399
v8,5,4,220.509114
v8,5,5,222.529699
v8,5,6,224.236719
v8,6,1,199.296811
v8,6,2,192.199622
v8,6,3,196.138832
v8,6,4,197.790249
v8,6,5,197.283257
v8,6,6,198.136669
v8,7,1,110.023948
v8,7,2,111.809473
v8,7,3,113.334062
v8,7,4,111.040694
v8,7,5,110.979502
v8,7,6,113.120494
v8,8,1,640.786958
v8,8,2,640.502172
v8,8,3,640.414708
v8,8,4,638.864723
v8,8,5,636.939979
v8,8,6,639.068674
head, v6 and v8.png (image/png, 15.5K) downloadview image
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-hackers@postgresql.org
Cc: ronan.dunklau@aiven.io, dgrowleyml@gmail.com, john.naylor@enterprisedb.com, jtc331@gmail.com, pgsql-hackers@lists.postgresql.org, ranier.vf@gmail.com, dilipbalaut@gmail.com
Subject: Re: [PATCH] Use optimized single-datum tuplesort in ExecSort
In-Reply-To: <2502562.vDctQ7PNa3@aivenronan>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox