[FFmpeg-devel] hwaccel infrastructure in libavcodec

Gregor Riepl onitake
Wed Mar 16 16:40:30 CET 2011

> I'll make a note that you are convinced VDA will neither be deprecated nor
> changed ;-)
> (Don't misunderstand me: It would be great if MPlayer/FFmpeg would support VDA,
> I am just surprised that everybody seems to believe the arguments that made XBMC
> not support VDA are invalid / weak.)

Well, at least we can flame Apple for changing it without notice. If
they suddenly change VideoToolBox, it would be a told you so...

But I do agree, one is never safe from API changes. I will check up on
the xbmc lists to see what their arguments were.

>> Well, for one, I don't think there is any reason why VDPAU deinterlacing
>> should not be offered by ffmpeg.
> I do not think this would work (the way you seem to expect it works):

Looks like I'm being too naive here.
That's why I'm asking for help from all experienced developers here,
after all. :)
Can you elaborate where the problem lies in my suggestion?

> (Apart from the fact that I still do not believe that it supports 60fps, but
> this doesn't matter atm)
> Of course, "playing" 1080 at 30 works fine with Ion, but did you test *decoding*?
> (I didn't, so it may well work fine.)

Ok, it looks like I used the wrong wording... It may very well be that
transferring decoded frames from video memory back to RAM would kill all
the performance benefits.
On the other hand, as is the case with VDA, hardware decoding
acceleration is mainly intended to offload playback to the GPU, instead
of pushing around full video frames in RAM and through the CPU.
But is hwaccel really designed to do decoding, instead of playback? The
device specific (virtual) pixfmts exist exactly for this reason, I believe.

More information about the ffmpeg-devel mailing list