Horton

Re: Improve CRC32C performance on SSE4.2

Andy Fan <zhihuifan1213@163.com>

View thread
From: Andy Fan <zhihuifan1213@163.com>
To: Jesper Pedersen <jesperpedersen.db@gmail.com>
Cc: Nathan Bossart <nathandbossart@gmail.com>, Tomas Vondra <tomas@vondra.me>, John Naylor <johncnaylorls@gmail.com>, "Devulapalli, Raghuveer" <raghuveer.devulapalli@intel.com>, "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>, "Shankaran, Akash" <akash.shankaran@intel.com>, Tomas Vondra <tomas@vondra.me>
Date: 2025-06-15T01:31:43Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Include _mm512_zextsi128_si512() in AVX-512 configure probes.

  2. Properly fix AVX-512 CRC calculation bug

  3. Workaround code generation bug in clang

  4. Compute CRC32C using AVX-512 instructions where available

  5. Inline CRC computation for small fixed-length input on x86

  6. Be more paranoid in configure's checks for CRC and POPCNT intrinsics.

Jesper Pedersen <jesperpedersen.db@gmail.com> writes:

Hi,

Thank you Nathan, Tomas and Jesper for the answers. The patch at [0]
works for me and I could work with master smoothly now.

> On 6/14/25 9:56 AM, Nathan Bossart wrote:
>> I'm also genuinely curious why folks are using -O0...
>> 
>
> Personally, I use
>
>  -O0 -fno-omit-frame-pointer
>
> for FlameGraph [1] investigations.

Same here. I use clang as compiler because I use clangd for code
indexing, gcc sometimes use different compiler options which may broke 
it. I used '-O0' in my daily coding and only use '-O2' when doing some
performance testing. 

[0]
https://www.postgresql.org/message-id/CANWCAZbsuavqUK4tg6UF-0-GVRMaq7BafUx4%2BDyd12y%3D-AuFAA%40mail.gmail.com 

-- 
Best Regards
Andy Fan