VP8 Over VDPAU In Gallium3D Is Emeric's Target

[smitty3268]

I imagine there is a debug mechanism in Mesa to do that already, not sure though.

This 1 line patch by Marek prints out compiled shader programs, although I'm not actually sure if that is TGSI or something else like classic Mesa IR.

https://bugs.freedesktop.org/attachment.cgi?id=45557

Example output:

https://bugs.freedesktop.org/attachment.cgi?id=45560

[runeks]

Looks like this didn't make it to the accepted projects ?

I can't find it in this list at least: http://www.google-melange.com/gsoc/p...oogle/gsoc2011

[mattst88]

That's really disappointing. I thought his proposal was very good and had an excellent chance of producing something quite nice.

[runeks]

Yep! It would be interesting if Google gave Emeric some sort of response as to why his project didn't get approved. They have approved these kinds of projects before.
But maybe that's the thing, they've learned how much it takes to actually implement a functioning GPU accelerated video decoder, and they don't think it's possible to do in 12 weeks?

There was the MPEG-2 XvMC project that didn't seem to finish according to the blog.

And I've also read about another guy who produced some code for XBMC to enable H.264 acceleration via GLSL, but that didn't produce anything production ready either, even though someone else picked it up later and worked on it some more (here).

Would definitely be interesting to hear Google's reason for not accepting this project. It would also help future students proposing projects of this kind to know what they should do to get their project accepted.

[mattst88]

Yep! It would be interesting if Google gave Emeric some sort of response as to why his project didn't get approved. They have approved these kinds of projects before.
But maybe that's the thing, they've learned how much it takes to actually implement a functioning GPU accelerated video decoder, and they don't think it's possible to do in 12 weeks?

There was the MPEG-2 XvMC project that didn't seem to finish according to the blog.

And I've also read about another guy who produced some code for XBMC to enable H.264 acceleration via GLSL, but that didn't produce anything production ready either, even though someone else picked it up later and worked on it some more (here).

Would definitely be interesting to hear Google's reason for not accepting this project. It would also help future students proposing projects of this kind to know what they should do to get their project accepted.

It's not Google's decision. It's X.Org's.

Apparently no mentor volunteered.

I think there was a big lack of communication.

[runeks]

Oh, that's really a shame then. H.264 acceleration is on the X.Org wiki SummerOfCodeIdeas. There's no proposal for a VP8-accelerating state tracker though.
Also, it says that the student should continue with the aforementioned g3dvl-code (MPEG-2 XvMC), but that didn't seem to be Emeric's idea of it on the mesa mailing list, as far as I've understood.