| From: | Nathan Bossart <nathandbossart(at)gmail(dot)com> |
|---|---|
| To: | "Amonson, Paul D" <paul(dot)d(dot)amonson(at)intel(dot)com> |
| Cc: | Bruce Momjian <bruce(at)momjian(dot)us>, Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>, Andres Freund <andres(at)anarazel(dot)de>, "pgsql-hackers(at)lists(dot)postgresql(dot)org" <pgsql-hackers(at)lists(dot)postgresql(dot)org>, "Shankaran, Akash" <akash(dot)shankaran(at)intel(dot)com> |
| Subject: | Re: Proposal for Updating CRC32C with AVX-512 Algorithm. |
| Date: | 2024-08-26 18:50:00 |
| Message-ID: | ZszOWAG5S6O8b9yD@nathan |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Mon, Aug 26, 2024 at 06:44:55PM +0000, Amonson, Paul D wrote:
>> I'm curious about where exactly the regression is coming from. Is it possible
>> that your build for the SSE 4.2 tests was using it unconditionally, i.e.,
>> optimizing away the function pointer?
>
> I am calling the SSE 4.2 implementation directly; I am not even building
> the pg_sse42_*_choose.c file with the AVX512 choice. As best I can tell
> there is one extra function call and one extra int64 conditional test
> when bytes are <256 and a of course a JMP instruction to skip the AVX512
> implementation.
And this still shows the ~14% regression in your original post?
--
nathan
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Amonson, Paul D | 2024-08-26 18:54:58 | RE: Proposal for Updating CRC32C with AVX-512 Algorithm. |
| Previous Message | Amonson, Paul D | 2024-08-26 18:44:55 | RE: Proposal for Updating CRC32C with AVX-512 Algorithm. |