If this is your first visit, be sure to
check out the FAQ by clicking the
link above. You may have to register
before you can post: click the register link above to proceed. To start viewing messages,
select the forum that you want to visit from the selection below.
No announcement yet.
AMD's UVD2-based XvBA Finally Does Something On Linux
Basically vaapi is only "working" when you want so when you use mplayer, vlc works from time to time too, currently it is broken with xvba, but mostly working with nvidia. Wrappers seem to be always problematic, the many issues i had with vdpau-video too. I can no say something about psb, that would have got native vaapi (g45 too, but currently only mpeg2, thats boring).
It's only problematic when the user can't seem to be able to install the right package or he always mixes up his installation on some weird distribution... Otherwise, it's working the same. In the XvBA case, it's even the only way to get useful things since by default this API (and implementation) is not that helpful, to remain "positive"...
I am really excited about all of this. While I fully appreciate bridgman's comments that AMD/ATi didn't promise much, I bought my 780G over a year ago and expected the UVD on it to work in the near future. I also mistakenly believed that all information about the cards had been publically released (including the UVD parts) and that Free drivers would be able to support this feature. Needless to say I was a bit disappointed when I found out the truth, especially as an nVidia alternative would have allowed VDPAU from the time that I bought it. I'm happy to see AMD/ATi using VA-API for its acceleration instead of creating yet another API.
On that note, I would really appreciate it if somebody who has the UVD/VA-API setup working could test that the samples on the following page are being properly accelerated: http://www.geekzone.co.nz/Fossie/4877
Thank you to everybody who is working on this.
Bridgman: I follow your posts avidly. They are very informative and helpful. I understand that AMD has not yet decided whether or not it will release the UVD specs to those developing Free drivers. This is really important to me — if I wanted to be stuck with non-free drivers, I would be using nVidia. The best result would be if the Free drivers could utilise the UVD hardware. If this isn't possible with the current architecture for DRM reasons, please put pressure on AMD/ATi to resolve this in future UVD versions so that these two things can be separated. I realise this thread is about the proprietary AMD/ATi drivers, but I want to flag that I only consider the non-Free drivers a stop-gap until the Free drivers accelerate video enough (whether through UVD, Gallium3D shaders etc.) to allow my CPU to decode 1080i in real time.
Aaron, I was unsuccessful in playing the freeview HD sample from here:
Once the Gallium3D driver and LLVM-compiled SW TCL starts working on your hardware you might be pleasantly surprised.
One of the biggest differences between the Catalyst driver and open source drivers on your chip was the software TCL code (aka running vertex shaders on the CPU). That code was pretty slow on classic mesa but apparently the LLVM compiler generates very efficient code, so hopefully you should see open source performance that is closer to Catalyst than what you have seen previously.
Don't think SW TCL is working yet on the Gallium3D driver but keep your eyes open for it.
Meh, I don't believe in Nvidia I am thinking I am going to buy a Radeon HD 5450.
XvBA does not support Radeon HD 5xxx. And last time I asked, it was not planned to be fixed. So, if some developer has some spare time and is really willing to dig into this code, this could be interesting for users. Thanks. ;-)