Re: Performance degradation on concurrent COPY into a single relation in PG16. - Mailing list pgsql-hackers

From John Naylor
Subject Re: Performance degradation on concurrent COPY into a single relation in PG16.
Date
Msg-id CAFBsxsF7GdxNbEeiRpTBBuiDBnyMH6d_HMehtZUZYZmdP2Tcqg@mail.gmail.com
Whole thread Raw
In response to Re: Performance degradation on concurrent COPY into a single relation in PG16.  (David Rowley <dgrowleyml@gmail.com>)
Responses Re: Performance degradation on concurrent COPY into a single relation in PG16.
List pgsql-hackers

On Thu, Jul 27, 2023 at 7:17 AM David Rowley <dgrowleyml@gmail.com> wrote:
>
> It would be really good if someone with another a newish intel CPU
> could test this too.

I ran the lotsaints test from last email on an i7-10750H (~3 years old) and got these results (gcc 13.1 , turbo off):

REL_15_STABLE:
latency average = 956.453 ms
latency stddev = 4.854 ms

REL_16_STABLE @ 695f5deb7902 (28-JUL-2023):
latency average = 999.354 ms
latency stddev = 3.611 ms

master @ 39055cb4cc (31-JUL-2023):
latency average = 995.310 ms
latency stddev = 5.176 ms

master + revert c1308ce2d (the replace-palloc0 fix)
latency average = 1080.909 ms
latency stddev = 8.707 ms

master + pg_strtoint_fastpath1.patch
latency average = 938.146 ms
latency stddev = 9.354 ms

master + pg_strtoint_fastpath2.patch
latency average = 902.808 ms
latency stddev = 3.957 ms

For me, PG16 seems to regress from PG15, and the second patch seems faster than the first. 

--
John Naylor
EDB: http://www.enterprisedb.com

pgsql-hackers by date:

Previous
From: Amit Kapila
Date:
Subject: Re: logical decoding and replication of sequences, take 2
Next
From: Bharath Rupireddy
Date:
Subject: Re: Adding a LogicalRepWorker type field