Horton

Re: Popcount optimization using AVX512

Nathan Bossart <nathandbossart@gmail.com>

View thread
From: Nathan Bossart <nathandbossart@gmail.com>
To: "Amonson, Paul D" <paul.d.amonson@intel.com>
Cc: David Rowley <dgrowleyml@gmail.com>, Andres Freund <andres@anarazel.de>, Alvaro Herrera <alvherre@alvh.no-ip.org>, "Shankaran, Akash" <akash.shankaran@intel.com>, Noah Misch <noah@leadboat.com>, Tom Lane <tgl@sss.pgh.pa.us>, Matthias van de Meent <boekewurm+postgres@gmail.com>, "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>
Date: 2024-03-18T17:30:04Z
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.

Attachments

On Mon, Mar 18, 2024 at 11:20:18AM -0500, Nathan Bossart wrote:
> I don't think David was suggesting that we need to remove the runtime
> checks for AVX512.  IIUC he was pointing out that most of the performance
> gain is from removing the function call overhead, which your v8-0002 patch
> already does for the proposed AVX512 code.  We can apply a similar
> optimization for systems without AVX512 by inlining the code for
> pg_popcount64() and pg_popcount32().

Here is a more fleshed-out version of what I believe David is proposing.
On my machine, the gains aren't quite as impressive (~8.8s to ~5.2s for the
test_popcount benchmark).  I assume this is because this patch turns
pg_popcount() into a function pointer, which is what the AVX512 patches do,
too.  I left out the 32-bit section from pg_popcount_fast(), but I'll admit
that I'm not yet 100% sure that we can assume we're on a 64-bit system
there.

IMHO this work is arguably a prerequisite for the AVX512 work, as turning
pg_popcount() into a function pointer will likely regress performance for
folks on systems without AVX512 otherwise.

-- 
Nathan Bossart
Amazon Web Services: https://aws.amazon.com