Page 3 of 4 FirstFirst 1234 LastLast
Results 21 to 30 of 39

Thread: Is AMD's New 2D Acceleration Architecture Still Slow?

  1. #21
    Join Date
    Dec 2008
    Posts
    988

    Default

    Quote Originally Posted by jrch2k8 View Post
    the issues is not the source code but the lack of the windows feature available to be accesed from a linux perspective and that is for both nvidia and amd(amd seems to have some libs here some apps there but nothing too proffesional either)

    what i think is needed is an official sdk of the internals gpu stuff like temperature, 3d options, 2d options if any, crossfire, fan speed, etc. that way any desktop enviroment can have a real control center with at least the important functions common to windows.
    I'm not sure what functionality from the Linux Catalyst Control Center is missing that is in the Windows version, or why anyone would want that. IMHO the Linux Catalyst Control Center is already cluttered enough.

    But AMD already has an SDK for stuff like GPU temp and more. So you can download the ADL SDK and build your own app with it. It has to be closed source though as the license prohibits the use of the SDK in open source programs.

  2. #22
    Join Date
    Jun 2010
    Location
    ฿ 16LDJ6Hrd1oN3nCoFL7BypHSEYL84ca1JR
    Posts
    1,052

    Default

    Quote Originally Posted by FunkyRider View Post
    I pity you all Catalyst users, you don't have no idea of what a perfect composite desktop from nVidia feels like. All the things you ever dreamed about in Linux desktop are already there several years ago. Just get yourself a piece of the lowest eng 8400GS and prepare to be blown away by the quality of nVidia driver.
    Don't feed the troll:
    http://www.phoronix.com/forums/searc...induser&u=6032

    For many user the 2d acceleration is buggy. Especially firefox has often problems with parts of the window going black.
    http://img715.imageshack.us/img715/5...41366x768s.png

    In the Archlinux Forums many have noticed it and the workaround is reverting to XAA.
    Code:
    aticonfig --set-pcs-str=DDX,ForceXAA,TRUE
    Someone who uses debian and had the same problem posted this.
    He has read it here in the phoronix forums:
    http://www.phoronix.com/forums/showthread.php?t=24323
    Seems to be the same problem.


    Strange thing is that d2d did work flawless when enabled with 10.5.
    Tested with xorg 1.7, xorg 1.8 patched to identify as 1.7, kernel 2.6.33 and 2.6.34. All the same, compositing on and off. All the same.

  3. #23

    Default

    Quote Originally Posted by not.sure View Post
    You mean "sucked away", because nvidia and 2D sucks badly. And "several years ago" it was even worse.
    This is incorrect. Just look at the recent benchmarks. Nvidia 2D rendering blows everything else out of the water. It used to be slow, but starting with the 180.xx series they began making massive improvements.

  4. #24
    Join Date
    Jan 2008
    Location
    South Africa
    Posts
    233

    Default

    Quote Originally Posted by thefirstm View Post
    This is incorrect. Just look at the recent benchmarks. Nvidia 2D rendering blows everything else out of the water. It used to be slow, but starting with the 180.xx series they began making massive improvements.
    But I think that is one of the primary points here. Benchmarks that have no correlation to how we experience it, cannot be used as proof.

    I currently have a Nvidia Mobile 9600gt in my notebook, and very often wish 2D was as fast as that cheap crappy little Intel IGP notebook I got my mom.

    So irrespective of how fast benchmarks run, the NVIDIA driver introduces very noticeable latency over one of the Open-Source solutions.

    What I can say, is that my Desktop with its Radeon 4850, which used to feel even slower than my notebook, now feels a slight bit faster. (which it should, the machine is easily 3 times more powerful)

  5. #25

    Default

    Quote Originally Posted by grigi View Post
    But I think that is one of the primary points here. Benchmarks that have no correlation to how we experience it, cannot be used as proof.

    I currently have a Nvidia Mobile 9600gt in my notebook, and very often wish 2D was as fast as that cheap crappy little Intel IGP notebook I got my mom.

    So irrespective of how fast benchmarks run, the NVIDIA driver introduces very noticeable latency over one of the Open-Source solutions.

    What I can say, is that my Desktop with its Radeon 4850, which used to feel even slower than my notebook, now feels a slight bit faster. (which it should, the machine is easily 3 times more powerful)
    Perhaps you should complain about this on the NvNews Linux forums. Many people including myself have very fast (as in usability) 2D with the Nvidia Linux drivers.

  6. #26
    Join Date
    Jun 2009
    Posts
    1,136

    Default

    Quote Originally Posted by monraaf View Post
    I'm not sure what functionality from the Linux Catalyst Control Center is missing that is in the Windows version, or why anyone would want that. IMHO the Linux Catalyst Control Center is already cluttered enough.

    But AMD already has an SDK for stuff like GPU temp and more. So you can download the ADL SDK and build your own app with it. It has to be closed source though as the license prohibits the use of the SDK in open source programs.
    well, a graphic overdrive panel is always more helpful, crossfire, video settings is kinda cool too, xvba settings. dont get me wrong i dont mean it for me, i can perfectly use an CLI interface but at bussiness level it looks strange to have the same driver in both oses and dont have the same options when possible ofc, well i guess there are so many bugs to fix that this wont be important for a while anyway.

    yeah about ADL SDK for linux well....

  7. #27
    Join Date
    Feb 2010
    Posts
    39

    Default

    Tasks like minimizing windows and scrolling feels more fluid and faster... However, our synthetic 2D benchmarks are actually showing a performance drop in a few quantitative tests.
    The Phoronix Test Suite facilitated this testing.
    Not really, I would rather say Phoronix Test Suite FAIL.
    But it's not really Breaking News. Since long ago Devs complain that PTS is doing the wrong (irrelevant) 2D Benchmarks.
    Never thought it would be THAT hard to begin to test with better Benchmarks like cairo-trace etc.
    Reminds me a bit of GL-Gears Benchs + Flamewars.

  8. #28
    Join Date
    Nov 2008
    Location
    Germany
    Posts
    5,411

    Default

    its simple to benchmark 10.5 vs 10.6

    i only ask my self "do you want use 10.5 again?"


    LOL Hell NO! 10-6 do have bug's but 10-5 is a nightmare!

    benchmark complete 10-6 win ALL benchmarks. (because i do not wana nightmares again)

  9. #29
    Join Date
    Aug 2007
    Posts
    6,631

    Default

    Did you really see a driver that works with KDE 4 correctly with the new accelleration active? I really think to disable it by default, because it only causes problems with gtk2 apps. Maybe it was tested with gnome (hint: ubuntu) only.

  10. #30

    Default

    I use kde 4, and while there is some weird behavior from time to time, its more of a "that's weird" reaction rather then something that even annoys me. Like a small box in an application flashing black for a second.

    Ovarall I'm happy with 10.6

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
  •