Page 12 of 15 FirstFirst ... 21011121314 ... LastLast
Results 111 to 120 of 149

Thread: Another Look At The Latest Nouveau Gallium3D Driver

  1. #111
    Join Date
    Jul 2008
    Posts
    914

    Default

    Quote Originally Posted by BlackStar View Post
    That's logistically impossible. The three major vendors have something like 300 people working on their closed-source drivers.
    Thats only half true, first they work not together, so you have only 100 who works on one of them. in opensource espacialy with gallium3d, much work is done much state trackers which work with all the free drivers. And even with the old non-gallium3d drivers there was much hardware-independend code there which could use all drivers.
    Quote Originally Posted by BlackStar View Post
    The open-source drivers are developed by maybe 30-odd people, several of whom unpaid.
    you forget here one thing unpayed people who do this have another motivation instead of money and they know they do good stuff, so they are better motivated, that free software has ofter better quality we have seen in the past. Because they dont have deadlines etc, they dont need to make shortcuts to get stuff done and get then often unmaintable code. AND we see today that the quality (not the speed) of the free ati driver is better then the blob for linux.

    Quote Originally Posted by BlackStar View Post
    It might sound harsh but the open drivers will never catch up to the blobs as things stand. Which is fine, since what they need to do is become good enough for the average user.
    we have TODAY first benchmarks in fullhd of games where the open driver is faster then the blob from amd. The point is to replace the blob, why would we do this in the first place? Linux is coming maybe not in 5-10 months but each year a bit, its a other question lets say the open driver gets support for accellerated video, is maybe in newer games not that fast, but is in any other way faster and better then the proprietary one, why would amd want then split most of there developers to a unstable nearly unusable binary blob that is not included in the kernel and for most distros in the latest stabe not usable? It would be a resource wastage to do that. Nobody with a clear mind did install fglrx on a r300-r500 card because they all lost more that they get (even if freedom is not the question) then amd stopped supporting it further with the closed drivers. That will happen each time faster with the newer chips, too.

    The big problem was/is that we had to reinvant nearly the whole X-core after the fork from Xfree86 and a long time with no free drivers, but this environment is now here and getting even better. The progress is very good on the free driver, the bazaar wins in long term always against the catedral.

  2. #112
    Join Date
    Jan 2011
    Posts
    220

    Default

    Quote Originally Posted by blackiwid View Post
    I donīt agree with you here, the free driver should in mid-long time replace the proprietary drivers. So like in cheap linux routers they provide then much proffesional features that not everybody needs. But thats not neccecary the work that amd have to do but they have to provide the documents that people need to make a driver with no restriction, also if amd give no uvd details, there are other ways to accellerate video other gpu, that goal has to be possible. Because else in long run the driver not even is good enough for home-users. I fancied to arm/chips because they seem to provide long batterie livetime and video-acceleration, but 3d and this video-acceleration is not in a free driver. So Arm is dead for me now. Because in 2-3 years or even faster? I will can play videos with 1080p in linux (without a 8core cpu) with high enough fps. Or then someday maybe 2560p or something like that.
    That had to be reachable else I revert my viewpoint and wait till someone provides me with that possibillity in a free driver. Dont get me wrong I do not demand it today to have it now, but I want in someday in the future, because else that specs or driver would not be free, its then not more or less free than the nv-driver because you only get a part that allows you to do x y but not z.

    Is linux that inefficient at video playback that it needs to have video excelleration ??

  3. #113
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,587

    Default

    Quote Originally Posted by Thatguy View Post
    Is linux that inefficient at video playback that it needs to have video excelleration ??
    Any OS is that inefficient without video decoding acceleration.

  4. #114
    Join Date
    Jan 2011
    Posts
    220

    Default

    Quote Originally Posted by deanjo View Post
    Any OS is that inefficient without video decoding acceleration.


    Seriously ?

    I mean I can decode mp4 1080p just fine with 50% of one 3.2 ghz cpu core no problem.










    trust me it was tough trying to keep the screenshot utility from driving up cpu useage. I'd say it averaged around 30-40% on a single core.

    Maybes its a linux problem ? I get similar results on windows.

    I am gonna try h264 next.

  5. #115
    Join Date
    Jan 2011
    Posts
    220

    Default

    I get the same result with h264

    minimal CPu usegage system idling watching a video cpu usegae is about 10-155 for both, with a few apps open its around 30%. screencapture really pumps the cpu useage up when you snap the shot.





    Uploaded with ImageShack.us

  6. #116
    Join Date
    Jan 2011
    Posts
    220

    Default

    this is with mp4 mp3 5.1 surround. just in case dumpt.com doesn't come back up, links on my above reply.

    1 core and multicore testing all give the same result. very limited cpu useage.







    Uploaded with ImageShack.us

  7. #117
    Join Date
    Jan 2011
    Posts
    220

    Default

    here ya go, 4 of the same h264 video being decoded at the same time, using only a vesa 2.0 renderer.

    Not even a 2d accelerating driver. Just vesa and software

    Granted it pushs the cpu pretty damn hard "running on one core" but it proves that it is easily doable.



    Uploaded with ImageShack.us

  8. #118
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,587

    Default

    Well a couple of things here. First of all BBB is not very demanding on bitrate, it is pretty easy on it in fact. You could try something that is a bit more demanding such as http://www.bealecorner.org/red/PFX_c...H.264-1080.mov (downscaled RED camera footage).

    Second of all you are still pushing your processor to ramp up to it's fastest p-state. With video decode acceleration on a processor like yours it shouldn't even hit 2% on a system such as yours on one core in its power save state (which is what my 4200+ locked in @ 800 Mhz).

    An atom for example is easily able to do high bitrate 1080p with proper video decode acceleration as well as some post processing as well (which is commonly used with vdpau on XBMC and such) while rarely hitting maybe 15% on a single core.

    I'm not saying you couldn't muscle your way though without video decode acceleration with a beefy enough processor but it certainly consumes a hell of a lot more resources especially when you are doing things like "scrubbing" through HD footage while editing.

  9. #119
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,587

    Default

    One other thing, try decoding at fullscreen (instead of scaling down)and watch your CPU usage.

  10. #120
    Join Date
    May 2007
    Location
    Third Rock from the Sun
    Posts
    6,587

    Default

    Grrr edit limit, it also looks like you are running at only 1280x1024 on your screen res.

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •