[pan.do/ra-users] parallel processing for videos

[j at pad.ma]

On 07/28/2013 06:10 AM, Nagarjuna G. wrote:
> Is there any specific reason?  In case we want to use opencl, is there
> a specific approach to follow?  Both nvdia and amd have GPU cores
> which are not expensive.  Would we be gaining advantage if we make use
> of this hardware?

there was some work on GPU accelerated x264 encoding and using that in
handbrake on os x, that might be something to look at, not sure this is
available on Linux but might be a start.
http://www.anandtech.com/show/5835/testing-opencl-accelerated-handbrakex264-with-amds-trinity-apu

> We have a friend who is willing to write code for it.  I wanted to
> know your opinion on this.

If he wants to work on it, to be useful for pan.do/ra transcoding,
looking at ways to accelerate VP8 encoding/decoding would be useful.

Possibly even VP9 as thats almost ready now. This would have to be in
the form of patches to libav/ffmpeg i.e. for scaling/color space
conversion or to libvpx for the actual encoding.

Looking around a bit I found
https://github.com/awatry/libvpx.opencl listed at
http://wiki.webmproject.org/vp8-implementations that uses opencl for
decoding.
http://comments.gmane.org/gmane.comp.multimedia.webm.devel/1503