[Ffmpeg-devel] [REQUEST] MMX/MMX2 and SSE optimizations for H.264 decoding

Guillaume POIRIER poirierg
Wed Sep 14 15:54:59 CEST 2005


On 9/14/05, Gamester17 <gamester17 at xboxmediacenter.com> wrote:
> Hi, I'm not a developer myself but I like to make a suggestion if that's
> possible?
> I have a Intel PIII CPU and I like to request MMX/MMX2 and SSE optimizations
> for H.264/AVC decoding on behalf of everyone with a x86 processor that
> supports MMX/MMX2 and SSE; Please, if you are a programmer with C and
> Assembly experience then please consider programming some MMX, MMX2 and/or
> SSE optimization to the H.264/AVC decoder (and demuxer) in
> libavcodec/libavformat (FFmpeg) for decoding on x86 processors (like Intel
> Pentium III x86 CPU in desktops and laptops), ...that way people with a
> slower processor can also enjoy smooth playback of H.264/AVC files in full
> quality, (and since many open source multimedia projects use FFmpeg as their
> main decoder library this will benefit a lot of people).


FFmpeg already features quite optimized CPU-hungry routines. I've
heard that even Nero's decoder is not as fast as FFmpeg. It could
still get even faster, but it would probably require lots of time for
small gains, or writing routines for a specific "core" of a CPU.

If your CPU is too slow, too bad for you. H.264 is a very hungry
codec. You can, however, disable some "eye candy" routines at decode
time like the inloop filter or things like that. MPlayer allows you to
control this with the option "-lavdopts" if I'm not mistaken (read the
man page).

Reading doesn't hurt, really!
  -- Dominik 'Rathann' Mierzejewski

More information about the ffmpeg-devel mailing list