Horton

Re: refactor architecture-specific popcount code

Nathan Bossart <nathandbossart@gmail.com>

View thread View on postgr.es
From: Nathan Bossart <nathandbossart@gmail.com>
To: John Naylor <johncnaylorls@gmail.com>
Cc: Heikki Linnakangas <hlinnaka@iki.fi>, pgsql-hackers@postgresql.org
Date: 2026-02-02T22:51:54Z
Lists: pgsql-hackers 
On Mon, Feb 02, 2026 at 09:16:42PM +0700, John Naylor wrote:
> It might be a good idea to do a little new testing, and I see a use
> for a special 8-byte path independent of AVX512: v6 seems to regress a
> little for single-words. But, it turns out that when gcc turns
> __builtin_popcountl into a single instruction, it's inline, but if it
> emits portable bitwise ops, it does so in a function called
> __popcountdi2(). That can be avoided by hand-coding in C for normal
> builds (and for 32-bit looks cleaner anyway), as in the attached 0005.

Oh, interesting.  I looked into this a little more [0].  Both gcc and clang
generate cnt instructions for aarch64, so we're good there.  However, clang
on x86-64 generates the bit-twiddling version, and gcc on x86-64 generates
a call to __popcountdi2() (which I imagine does something similar).  It's
not until you provide a compiler flag like -march=x86-64-v2 that gcc/clang
start generating popcnt instructions for x86-64, which makes sense.  0005
seems like the correct move to me...

[0] https://godbolt.org/z/he3WozG3E

-- 
nathan

Commits

  1. Make use of pg_popcount() in more places.

  2. Remove uses of popcount builtins.

  3. Remove some unnecessary optimizations in popcount code.

  4. Remove specialized word-length popcount implementations.

  5. Move x86-64-specific popcount code to pg_popcount_x86.c.

  6. Refactor some SIMD and popcount macros.

  7. Rename "fast" and "slow" popcount functions.

  8. Rename pg_popcount_avx512.c to pg_popcount_x86.c.

  9. Remove trailing zero words from Bitmapsets

  10. Don't use _BitScanForward64/_BitScanReverse64 on 32-bit MSVC builds