PDA

View Full Version : An Article on VAAPI?


sreyan
12-04-2007, 11:12 PM
Hi Michael,
Looking on the web, i've found that intel is working on a new video acceleration api, vaapi. While the software is still in the planning phase and article on vaapi would be very interesting. This maybe premature however
Thanks

Sreyan

inewbie
01-04-2008, 12:29 AM
iu am very interesting in this too.

Regenwald
01-05-2008, 05:26 PM
well, if you'll write about vaapi, i think that it is necessary to explain gallium3d, dri2, Input Redirection, Redirected Direct Rendering, llvm, glucose, ttm, mpx and so on, too. we're going to see several improvements on the x-server which are really exciting...

lakritz
01-05-2008, 07:28 PM
well, if you'll write about vaapi, i think that it is necessary to explain gallium3d, dri2, Input Redirection, Redirected Direct Rendering, llvm, glucose, ttm, mpx and so on, too. we're going to see several improvements on the x-server which are really exciting...

Please do, that would be really interesting.

Regenwald
01-07-2008, 04:33 PM
well, michael is at ces at the moment. we have to remember him when he's back...:)

Gamester17
02-20-2008, 10:03 AM
Still no news article on VaAPI from Phoronix.com?

I only found the news article on "A Unified GPGPU API In Gallium3D"
http://www.phoronix.com/scan.php?page=news_item&px=NjMyMg

Maybe VaAPI could be implemented in Mesa but above the Gallium3D architecture in a similar manor to the one suggested for GPGPU in Zack Rusin's blog; though as a unified hardware assisted video decoding API above Gallium3D written using a state tracker?

Zack adds that as Gallium3D now uses LLVM (Low Level Virtual Machine) for compiling and accelerating shaders, any programming language could be put on the GPU with "minimal efforthttp://zrusin.blogspot.com/2008/02/gpgpu.html

Then make VaAPI support mandatory in each Gallium3D based device driver :) ...maybe a good idea since VaAPI spport is required both in the video codec decoder and the device driver for the graphics hardware?

PS! a generic XvMC library ('libxvmc') would probably be implemented in a similar way for backwards compatibility for video codecs and device drivers that do not yet feature VaAPI support (which today is most if not all).

Hope that this new Video Acceleration API will recieve a little more attention from the press in the future :)