#520 [REQ] AVX512 optimizations

[forart.eu]

Hi everyone,
seems that AVX512 builds performs pretty well compared to "normal" x64 oones (~ x40 vs x350 according to this HA 3ad) but, as FFMPEG devs teaches, specific (manual ? LLM-driven ?) code optimizations can boost up results even more.

Does anyone plans to bring them in ?

Thanks in advance !

[Alexander Leidinger]

Some AVX2 optimizations are now in the development tree and will be part of 4.1.