Re: HD-Video Hardwar Acceleration?
On 29-8-2009 12:53, Harald Braumann wrote: I for one care less about 3D acceleration. But it's very frustrating that it's still not possible to watch full HD videos on Linux. This is slightly off-topic, but I had to comment on your statement. I have been watching 1080p content on my linux-htpc for over a year using Intel's G35-videochip and a E6750 (2.6Ghz Core 2), which could decode (almost?) all of the video content I threw at it with at most a minor stutter so now and then. My system did start stuttering badly on the famous bird scene, but there isn't much material that is that heavy. It would also start to stutter when there are heavy background jobs, but using 'nice' and 'ionice' on those jobs will help a lot. I switched to a VDPAU-enabled motherboard last week, and there is no significant stuttering with cpu-heavy background jobs running, although disk-heavy jobs still make it stuter (so I still need ionice). My mplayer dropped from 70-100% cpu to 0-2% during full-hd decoding. And it indeed requires much less power to play video. The new motherboard is so much more power-efficient, it now consumes less power during video-playback than the old one did when idling (now idle ~ 70W, full hd playback 75-80W vs, old idle ~ 80W and full hd ~ 100W). So your statement is not true. There are fast enough cpu's to do the decoding on their own and there is video-decoding available if you can live with the binary drivers from nvidia (at least for now). So there are two basic choices; you could pick-up a video-chip that will likely get video-decoding in the future (like the Intel G45 and better) and combine it with a fast cpu (like a E8200) to use for now, or pick up a motherboard with a nvidia chip (like the G9300/9400) and use the binary drivers untill the open source guys catch up. If you're more concerned with power-consumption, the latter is your best bet. If you're more concerned with not tainting your linux environment, you should go with the first. Best regards, Arjen ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
OT: Re: HD-Video Hardwar Acceleration?
On Mon, 31 Aug 2009, Arjen van der Meijden wrote: On 29-8-2009 12:53, Harald Braumann wrote: I for one care less about 3D acceleration. But it's very frustrating that it's still not possible to watch full HD videos on Linux. This is slightly off-topic, but I had to comment on your statement. I know I cured some audio stutters by getting the kernel-rt package (kernel with realtime patches) from the CCRMA project (I'm on Fedora). Maybe this would work with Video too. :) - | Name: Tim Nelson | Because the Creator is,| | E-mail: wayl...@wayland.id.au| I am | - BEGIN GEEK CODE BLOCK Version 3.12 GCS d+++ s+: a- C++$ U+++$ P+++$ L+++ E- W+ N+ w--- V- PE(+) Y+++ PGP-+++ R(+) !tv b++ DI D G+ e++ h! y- -END GEEK CODE BLOCK- ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
On Sat, Aug 29, 2009 at 6:53 AM, Harald Braumannha...@unheit.net wrote: Hi, is there any chip out there for which hardware accelerated HD video decoding is supported by an open-source driver? Or for which at least documentation exists so that it could be implemented? Video decode acceleration (at least motion compensation) for all radeon chips could be implemented with the currently available documentation. For r3xx and above it would be implemented in shaders. For r1xx/r2xx, the 3D engine has a special MC mode. Alex ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
On Sun, Aug 30, 2009 at 1:46 AM, Harald Braumannha...@unheit.net wrote: since we don't have documentation on the dedicated video decoders onboard most GPUs, and we haven't reverse-engineered them, So there is documentation on some? Which ones? I couldn't find any information about that. He said we _don't_ have documentation ... since we *don't* have documentation ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
You could give this a try. Thanks --Shuang Zou, Nanhai wrote: Hi, We have enabled VAAPI support for G45 and GM45 platform. I have tested it with some high profile 1080p mpeg2 stream. Using a patched mplayer at http://www.splitted-desktop.com/~gbeauchesne/mplayer-vaapi/ I can see CPU usage is dramatically reduced. Currently, we only support mpeg2 decoding. Welcome to test it. The git repo is at git://git.freedesktop.org/git/libva, To successful compile the mplayer, you need to choose the libva31 branch. Thanks Zou Nan hai ___ Intel-gfx mailing list intel-...@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/intel-gfx Harald Braumann wrote: Hi, is there any chip out there for which hardware accelerated HD video decoding is supported by an open-source driver? Or for which at least documentation exists so that it could be implemented? I for one care less about 3D acceleration. But it's very frustrating that it's still not possible to watch full HD videos on Linux. Cheers, harry ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
On Sat, 29 Aug 2009 19:27:29 +0800 Shuang He shuang...@intel.com wrote: You could give this a try. Thanks --Shuang Zou, Nanhai wrote: Hi, We have enabled VAAPI support for G45 and GM45 platform. I have tested it with some high profile 1080p mpeg2 stream. Using a patched mplayer at http://www.splitted-desktop.com/~gbeauchesne/mplayer-vaapi/ I can see CPU usage is dramatically reduced. Hi, but I would still need a vaapi implementation that utilises hardware acceleration. Cheers, harry signature.asc Description: PGP signature ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
On Saturday, 29th of August 2009 10:53:06+0200, Harald Braumann wrote: Hi, is there any chip out there for which hardware accelerated HD video decoding is supported by an open-source driver? Or for which at least documentation exists so that it could be implemented? I for one care less about 3D acceleration. But it's very frustrating that it's still not possible to watch full HD videos on Linux. I don't know if this guy is still working on it, but you might want to have a look at http://bitblitter.blogspot.com/2009/01/yes-im-still-decoding-video-using.html I don't know much about video decoding, but maybe the best way to do video acceleration is to implement the decoder in OpenCL, because it will work on future gallium3d drivers and (most important) the implementation could be completely open source. (One could also implement advanced deinterlacing and other video filters in an efficent way) Unfortunately nobody is working on this right now ... ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
On Sat, Aug 29, 2009 at 2:48 PM, cc@gmx.de wrote: On Saturday, 29th of August 2009 10:53:06+0200, Harald Braumann wrote: Hi, is there any chip out there for which hardware accelerated HD video decoding is supported by an open-source driver? Or for which at least documentation exists so that it could be implemented? I for one care less about 3D acceleration. But it's very frustrating that it's still not possible to watch full HD videos on Linux. I don't know if this guy is still working on it, but you might want to have a look at http://bitblitter.blogspot.com/2009/01/yes-im-still-decoding-video-using.html I don't know much about video decoding, but maybe the best way to do video acceleration is to implement the decoder in OpenCL, because it will work on future gallium3d drivers and (most important) the implementation could be completely open source. (One could also implement advanced deinterlacing and other video filters in an efficent way) Unfortunately nobody is working on this right now ... But using OpenCL instead of dedicated video hardware would result into patent problems so that many distros cannot ship it by default. ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
On Saturday, 29th of August 2009 13:03:58+0200, drago01 wrote: But using OpenCL instead of dedicated video hardware would result into patent problems so that many distros cannot ship it by default. Why should patent problems depend on the programming language? ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
On Sat, Aug 29, 2009 at 3:26 PM, cc@gmx.de wrote: On Saturday, 29th of August 2009 13:03:58+0200, drago01 wrote: But using OpenCL instead of dedicated video hardware would result into patent problems so that many distros cannot ship it by default. Why should patent problems depend on the programming language? They do not depend on the programming language but whether it is done in software or hardware. From a patent POV there is no difference between an opencl or a C implementation (you are implementing the patented algorithms), but if you are just talking to hardware decoding chip you don't. ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
On Sunday, 29th of August 2009 15:06:10 +0200, drago01 wrote: They do not depend on the programming language but whether it is done in software or hardware. From a patent POV there is no difference between an opencl or a C implementation (you are implementing the patented algorithms), but if you are just talking to hardware decoding chip you don't. Distributions ship lots of patented decoders right now, why should stop doing that, just because the decoder is implemented in OpenCL? I understand your concerns, but they are not really new. Plus, I would prefer an open source software solution, even if it violates (stupid) software patents, over closed hardware solutions. ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
I don't feel like digging through emails and writing point-by-point replies, so I'm just going to do point-by-point talking-to-myself. - The APIs for accelerating full decoding and not just colorspace conversion and scaling (Xv) are: -- VA-API -- VDPAU -- XvBA -- XvMC Of those, only VDPAU is worth implementing. XvMC can be done in terms of VDPAU if you're creative enough. - OpenCL would be fine for video, but stupid and slow, just like OpenGL-based video decoding. - The current plan is to *not* add any more video decoding to any DDX, but to do it in Gallium using a state tracker. Expand g3dvl to be a true pipeline, let drivers fill in the pipeline with whatever they can accelerate, and then expose VDPAU and XvMC to userspace. Of course, only radeon (me, MrCooper) and nouveau (marcheu, ymanton) have actually agreed that this is a good idea, but the intel devs will probably keep maintaining their XvMC and VA-API implementation either way. - It's not done because we've been working on other things. If it makes you feel better, 3D support *is* required for this kind of work, since we don't have documentation on the dedicated video decoders onboard most GPUs, and we haven't reverse-engineered them, so we'll have to use 3D shaders instead. ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
Re: HD-Video Hardwar Acceleration?
On Sat, 29 Aug 2009 15:55:09 -0700 Corbin Simpson mostawesomed...@gmail.com wrote: - It's not done because we've been working on other things. If it makes you feel better, 3D support *is* required for this kind of work, It neither makes me feel better nor worse. My mail could have been read as a complaint about people working on 3D rather than video acceleration. This was not my intention at all. Sorry, if it sounded that way. since we don't have documentation on the dedicated video decoders onboard most GPUs, and we haven't reverse-engineered them, So there is documentation on some? Which ones? I couldn't find any information about that. so we'll have to use 3D shaders instead. AMD's John Bridgman has said somewhere (sorry, don't have the reference handy), that maybe in the future the specifications of the UVD could be released to the public. Should that ever happened, would it be possible to use it from within that state tracker? Or would the state tracker be tied to tightly to the shader model? Cheers, harry ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg signature.asc Description: PGP signature ___ xorg mailing list xorg@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/xorg
