Hi Gwen,
first of thanks for your work at xvba-video. When mpeg-2/mpeg-4 is released I plan on integrating this into xvba-video after our xbmc is done. I did not see the updated SDK yet, but as these are bitstream codecs again. I think we can use our infrastructure, just the mapping of ffmpeg to xvba structs should be done.
If you think of grepping "libAMDXvBA.so.1", I did this multiple times and found: @HWUVD_H264Level51Support
But, yeah every time I tried to enable it through amdscdb I got a black screen.
Any other hint?
Bye
Peter
Announcement
Collapse
No announcement yet.
AMD Comments On XvBA Video Situation
Collapse
X
-
Originally posted by markg85 View Postand that is exactly why we're way better off using UVD directly. No more waiting till AMD decides to make it usable. Reverse engineering anyone? ^_-
Besides, the PCS key is pretty easy to find, this is what I did. The constraints and limitations of this support are available under NDA, though. Wait for Tim to disclose the details to you. I think he hinted enough now. At least, he disclosed there was a "PCS key". Just go find it out.
Leave a comment:
-
Originally posted by uid313 View PostRadeon 1xxx is older than my grandma.
Leave a comment:
-
Originally posted by uid313 View PostAMD has been open source for 5 years now, they started in 2007.
And the open source device driver still sucks?
Leave a comment:
-
Originally posted by uid313 View PostHurry up guys, make new XvBA update.
If porno don't work, I don't buy AMD.
Leave a comment:
-
hmm ok... informative comments in this thread... about the video-capabilities... yes it woudl be great to have some uvd like feature... I have here 2 zacate pcs... one htpc/server and one notebook... on both I get not always fluent results with 720p x264 video files... so that sucks a bit...
to redhat I am very glad that they do good job here especialy with the amd drivers... but fedora has some issues... I even wrote a bug report... so I cant use it ^^ it seems a response to such a bug report takes long... so maybe off-topic but maye a amd guy is here ^^
its a known bug btw, but seems to interest nobody at all for a long time...strange that its distribution specific thought that that should be happen very seldom because all use the same upstream...
but I should maybe debug it by myself, just trying other kernels other alsa versions... but I am not shure that I migrate form ubuntu to fedora anyway ^^ so I have no big interest in invest so much time ^^...
ok but that gets to off-topic...
is there any very ruff prognosis when some kind of video acceleration for the radeon driver could hit mesa or gallium3d?
I would be happy about that... then maybe better powermanagement but even that is not that important... because that kind of semiworks ^^ and have not very big issues with that... I dont cant think anything thats so important than this feature.... yes some guys need more 3d speed or better opencl support but at least the first of it would be way harder to get the driver to nearly the same speed as the closed ones fast... but uvd like feature would make so much people happy...
Leave a comment:
-
Originally posted by markg85 View PostYou're mostly right.
First, cairo has nothing to do with it
and 5 years to go to windows 7 comparable graphics.. Nope, sorry. The oss driver state is still quite horrible but it works and usually works better then the AMD drivers in terms of stability. However, in performance, power management and amount of noise produced it's still _far_ from being usable. But they are getting very close to good working drivers. I'd say probably about 2 more years before it's really in a good working shape.
If you want to compare it against windows then i'd say we are at the point where windows vista RC was released with the massive graphic driver issues
* you get me wrong here im not referering to an specific driver/GPU, im talking about theorical graphic stack capabilities and in this case linux new graphic stack is getting very close in features to win7 and in some cases is fairly superior
* well for my desktop PC user r600g is perfectly fine and 100% usable and well having an 1200W PSU is not like few watts more are going to ruin my day, aka
-- kwin with gl working uber
-- xonotic at ultra setting for clear the mind works peachy and fast enough at 60ish FPS
-- firefox/chromium webgl and render accel works very good
-- mplayer/ffmpeg MT and my phenom II X4 965 handle 1080p 60 5.1 like a champ with Xv, so im not desperated for UVD(i expect my AMD FX 6200 to improve it more after make world with gcc 4.7)
-- blender works peachy for my hobbists designs
-- Lineage2 Tauti/AION in wine 1.5.6 + r600g is peachy here too. i could use some more FPS in sieges but the engine in lineage2 sucks ass even on win7
so im happy with r600g in my specific use case scenario, so for now "it sucks" depending on your needs
Leave a comment:
-
Originally posted by uid313 View PostKernel and DRM was written by David Airlie.
Don't knot who wrote DRI but it was probably Keith Packard cuz hes awesome!
GEM was written by Intel. Then someone wrote TTM.
Wayland was written by Kristian H?gsberg.
Gallium3D was written by Tungsten Graphics who got bought by VMware.
Xorg was written by Keith Packard.
So what the fuck did AMD do all this time?
even so they have hired some good devs from the community to help, example the initial SI mesa patch among other stuff.
and obviously AMD provide support for the devs in case they have questions or doubts about internals of the GPU ASM API.
i agree in the fact they could help a bit more but is still way better than what nvidia do tho
Leave a comment:
Leave a comment: