[FFmpeg-devel] AVX assembly for ffmpeg

[Daniel Kang]

I have been asked to write AVX assembly for ffmpeg. This will require
porting x264's x86inc.asm file with the AVX abstraction layer. Ideally, I
would like to bump the yasm version, as it is already checked for pextrd
(SSE4). However, is this is not permissible, I can see if the AVX compile
flag can be set conditionally, based on the yasm version.

I have already ported many functions in x264, yielding substantial benefits.
This would macro functions, which is already done in
libavcodec/x86/h264_intrapred.asm. Are there any comments?