[FFmpeg-devel] [PATCH] avcodec/v210enc: add new function for avx2 avx512 avx512icl
James Darnley
jdarnley at obe.tv
Mon Oct 31 14:52:15 EET 2022
>> + %else
>> + pand m1, m6, m1
>> + pandn m0, m6, m0
>> + por m0, m0, m1
>> + %endif
>
> Isn't that pattern a vpblendb or some such ?
I think Kieran already responded to this on IRC but I will too.
Unfortunately not. This blend is at the bit level. This is v210 so the
packing has the middle sample overlapping with the bottom sample in the
second byte.
I also want to amend my performance numbers on Broadwell. I can confirm
Kieran's disagreement and can reproduce the 10% speed up on it:
1676±14.6 vs 1426±20.9
I will re-check Zen and amend the commit message as necessary.
More information about the ffmpeg-devel
mailing list