[FFmpeg-devel] cuda

Benjamin Zores ben
Fri Dec 28 11:54:22 CET 2007

Michael Niedermayer a ?crit :
> On Thu, Dec 27, 2007 at 11:37:01PM +0300, Anatoliy wrote:
>> Michael Niedermayer wrote:
>>> Noone tried AFAIK, but if you want to try patches are certainly welcome
>>> though likely just (I)DCT will not lead to much speedup.
>>> [...]
>> As I have seen after code reseaching there is one big structure 
>> DSPContext containing all available methods for DSP optimization. There 
>> are too much methods. Which of them mostly important for codec time 
>> optimization?
> I do not think that adding cuda support via DSPContext will work out very
> well. The reason is that these functions are all working on small amounts
> of data at a time ...

 From what I remember of CUDA design (and GeForce 8 h/w in general),
we're talking about 1000s of threads doing some very specific 
calculation which involves a design far different from DSPContext.
It could be seen like PS3' SPUs but at a much larger scale.

But again, as Rathann said, even if CUDA SDK is free (to download and 
use at least, not sure about open-source), it relies on nVidia 
proprietary driver for Xorg.

> XVMC IIRC supports just mpeg1/mpeg2 though the design can easily be extended
> to mpeg4 ASP (divx4/5) and msmpeg4v3 (divx3). H.264 though is a different
> beast and will require a seperate (and more complex) solution.

However the only extension I know about comes from VIA driver
but seems to be h/w specific which is bad imho.
XVMC so called successor seems to be VA-API but I've already started a 
thread about it months ago.


More information about the ffmpeg-devel mailing list