AMD's UVD2-based XvBA Finally Does Something On Linux

[energyman]

No, /usr/lib32 is a Debian/Ubuntu aberration

and gentoo..

btw, when will it possible to compile&install your stuff with a different libtool version than the one used by ubuntu?

[Kano]

@monraaf

The chart is not fully correct when you try to use the vdpau accellerated flash. Then it is possible to see vdpau caused corruptions (even gpu lockups with a bit older drivers) especially when you use Xserver 1.4 (Debian Lenny). You see the famous Xid errors in system log then, running a fullscreen 3d app helps to get rid of curruptions.

Also the sample picture is incorrect as this has been fixed with driver 10-12 for evergreen as long as you only want to decode h264 up to l4.1 or vc1. The only partital support for mpeg(1/2) has been dropped from xvba-video, only VLD for h264/vc1 is there anymore. So ati should fix l5.1 issues this year... A vdpau lib would be nice too or at least they should convince adobe to support vaapi...

[monraaf]

Also the sample picture is incorrect as this has been fixed with driver 10-12 for evergreen

Yeah it took them about a year to fix it I believe... Oh wait I can't say that, they never officially released UVD for the Linux Desktop in the first place, can't complain about bugs in unreleased products!

A vdpau lib would be nice too or at least they should convince adobe to support vaapi...

Now that would be a hard sell, considering that they don't even support vaapi themselves. Note that keeping UVD on the Linux desktop in a perpetual unreleased status and only allowing access to the UVD block by a third-party library and therefore shifting the support burden to the company behind the third-party library who can't do anything about the UVD related bugs that are present in fglrx is not what I consider as support.

[monraaf]

Hey, Evergreen support broke again in 11.1 drivers? So, it was fixed only for one to two Catalyst releases? I don't even want to try...

I have no idea if it's broken again, I have no access to 11.1 yet. This was just me playing with GIMP, sorry about that

[deencecit]

Adler

??????? ??????? ? ??????.
<a href=http://namore.tv>?????????? ?????? ?????</a>

[DanL]

Note that keeping UVD on the Linux desktop in a perpetual unreleased status and only allowing access to the UVD block by a third-party library and therefore shifting the support burden to the company behind the third-party library who can't do anything about the UVD related bugs that are present in fglrx is not what I consider as support.

ATI's official response to that:
Good support is like beauty, which means it's in the eye of the beholder.
P.S.: Don't bother sending us your r?sum?.

[popper]

ATI's official response to that:
Good support is like beauty, which means it's in the eye of the beholder.
P.S.: Don't bother sending us your r?sum?.

LOL well now that their CEO Dirk Meyer that thought it was a good idea and so sold off their manufacturing unit Doh, is gone after a third-quarter net loss of US$118 million.

http://www.goodgearguide.com.au/article/372970/meyer_resigns_amd_ceo/

the Executive Board are on the lookout for someone to save them from the Intel/ARM onslaught so perhaps they will take those r?sum? after all

Two Years and all change again, is it bad or not getting through two CEO's per new CPU design cycle

[Kano]

Well UVD is even the worst solution for flash 10.1 accelleration... Maybe ATI should even polish the win drivers Nvidia+Intel beat UVD with ease...

http://www.tomshardware.com/reviews/adobe-flash-10.1-performance-hardware-acceleration,2805.html

[evolution]

Well UVD is even the worst solution for flash 10.1 accelleration... Maybe ATI should even polish the win drivers Nvidia+Intel beat UVD with ease...

http://www.tomshardware.com/reviews/...tion,2805.html

I saw that comparision too...
It seems current Catalyst's UVD2 implementation isn't working very well on Windows too... (But I can't confirm it, since my 4650AGP card is installed with Windows XP).

Hope the new Catalyst version that's coming solves the video decoding problems on Windows/Linux OS's...

Cheers

[popper]

as does ARM flash 10.1 it seems, and they dont have much in the way general optimised SIMD code blocks acceleration or SIMD optimised app's come to that yet.

what is odd is that a hard-core engineer such as Dirk Meyer would screw it up so much, to many yes men around perhaps! selling off the core.

perhaps he should have got out the office more and listened to the end user chatter more

"http://www.brightsideofnews.com/news/2011/1/10/coup-at-amd-dirk-meyer-pushed-out.aspx One of our sources, a former executive within the company was less than kind in his texts: "Dirk touched everything into Gold at DEC. In AMD, everything after K7 turned into Sh*t." When asked to elaborate, we were told that "Dirk and Hector run the company to the ground by selling the family jewlery. Dirk did not believe in Cellphone business and gave everything to BroadCom and Qualcomm. We lost Nokia, Motorola, Sony Ericsson and look who is now taking all the deals."