[FFmpeg-devel] ffmpeg future in regards to hardware decoding & software filtering.

Genevill Jevi\\\'psyo royal.envy at gmail.com
Fri Mar 18 17:37:42 CET 2011

Gregor Riepl <onitake <at> gmail.com> writes:

> >> of course theres a chance, it just depends on a volunteer doing the work and
> >> it being license wise possible
> > 
> > I hop that some dev’s take interest in GPU accel, I know such a thing is a
> > massive undertaking though.  I was just curious because ffmpeg is one of the
> > most important software’s ever developed...
> OpenCL support is something I was eyeing with myself, but for now, I
> think it would make more sense to work on hwaccel in general first. If
> the interface for modules is simple enough, it will be much easier to
> add any kind of hardware support.
> Are you thinking of decoding specifically or using OpenCL for any
> operation where it would make sense to accelerate?
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel <at> ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Thank for your response!

Ok I come representing a community of videophiles on avsforum.org, those of us
with NV cards use CoreAVC CUDA to accelerate H.264 with the hardware VP while
still having the ability to preform software filter with ffdshow & AviSynth
plugin.  The AviSynth scrips we use for MVTools2 creates tremendous load on the
CPU so we use hardware acceleration to for decoding of the video stream to
reduce CPU load...

So this is the layout: MPC-HC (any directshow player) + CoreAVC CUDA + ffdshow
with AviSynth + madVR.

CoreAVC 2.0 works perfect for those with NV cards that support CUDA but there is
no solution for those with ATI cards, that why I was hoping ffmpeg would add
openCL support (or other) for GPU acceleration.  Please if you know of any way
this can be implemented let me know, we can use all the help we can get!...

Cheers, Genevill,

More information about the ffmpeg-devel mailing list