Horton

Re: Popcount optimization using AVX512

Ants Aasma <ants.aasma@cybertec.at>

View thread
From: Ants Aasma <ants.aasma@cybertec.at>
To: Nathan Bossart <nathandbossart@gmail.com>
Cc: Alvaro Herrera <alvherre@alvh.no-ip.org>, "Amonson, Paul D" <paul.d.amonson@intel.com>, Tom Lane <tgl@sss.pgh.pa.us>, David Rowley <dgrowleyml@gmail.com>, Andres Freund <andres@anarazel.de>, "Shankaran, Akash" <akash.shankaran@intel.com>, Noah Misch <noah@leadboat.com>, Matthias van de Meent <boekewurm+postgres@gmail.com>, "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>
Date: 2024-04-04T13:02:53Z
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. Fix __attribute__((target(...))) usage.

  2. Use __attribute__((target(...))) for AVX-512 support.

  3. Fix code for probing availability of AVX-512.

  4. Optimize visibilitymap_count() with AVX-512 instructions.

  5. Optimize pg_popcount() with AVX-512 instructions.

  6. Inline pg_popcount() for small buffers.

  7. Avoid function call overhead of pg_popcount() in syslogger.c.

  8. Refactor code for setting pg_popcount* function pointers.

  9. Inline pg_popcount{32,64} into pg_popcount().

  10. Remove MSVC scripts

  11. Use ARMv8 CRC instructions where available.

  12. Use Intel SSE 4.2 CRC instructions where available.

On Thu, 4 Apr 2024 at 01:50, Nathan Bossart <nathandbossart@gmail.com> wrote:
> If we can verify this approach won't cause segfaults and can stomach the
> regression between 8 and 16 bytes, I'd happily pivot to this approach so
> that we can avoid the function call dance that I have in v25.

The approach I posted does not rely on masking performing page fault
suppression. All loads are 64 byte aligned and always contain at least
one byte of the buffer and therefore are guaranteed to be within a
valid page.

I personally don't mind it being slower for the very small cases,
because when performance on those sizes really matters it makes much
more sense to shoot for an inlined version instead.

Speaking of which, what does bumping up the inlined version threshold
to 16 do with and without AVX-512 available? Linearly extrapolating
the 2 and 4 byte numbers it might just come ahead in both cases,
making the choice easy.

Regards,
Ants Aasma