Intel Considering VDPAU Support For X.Org Driver
Phoronix: Intel Considering VDPAU Support For X.Org Driver
In addition to learning that Intel's UXA acceleration architecture will live on in its current form, there were a few other interesting bits of news reported as well during the X.Org talks at FOSDEM 2009. During Eric Anholt's talk some video playback/decoding improvements were discussed. For its xf86-video-intel driver, Intel is still working on adding support for more codecs to XvMC along with possible support for offloading VLD to the GPU during MPEG-2 decoding...
Wise words, the power management thing is indeed an important point. But more important than power saving is that there *is* a video acceleration method. I'm interested into why they look more at VDPAU instead of their self developed and free VA API. Too bad that there is not one open soruce driver that supports the latter.
I'm looking forward to see the records.
Originally Posted by bugmenot
I think the reason for the interest in VDPAU is simply a case of where because of it's rapid development and adoption it has basically set it's self as the standard. Given it's short maturation time it would probably not be worth the time nor effort to develop their own API when a present working solution can be adapted just as easily.
So, are there any news considering VDPAU support in intel drivers.
I think it would be the best to add support for VDPAU, because the most important application already have support for it.
Power management Intel, just say so, we all know the main reason you dislike shaders is that your HW does not have powerful shaders.
+1 to "NOT using shaders for video decoding is ridiculous".
In the recent years we've seen a perfect example of generic shader hardware overtaking the role of specific hardware, specifically the 2D operations on R600 (I think) and newer - the 2D hardware was scrapped, since the shader hardware became so fast that having extra hardware for 2D was a waste of silicon. I don't know much about nvidia chips but I suspect they reached a similar conclusion.
If I put this next to the long-running trend of moving hardware functions into software (drivers - winmodems, 80211mac, anyone?), this is not new at all.