Results 1 to 9 of 9

Thread: GLAMOR 0.5 Delivers Performance Improvements

  1. #1
    Join Date
    Jan 2007
    Posts
    13,405

    Default GLAMOR 0.5 Delivers Performance Improvements

    Phoronix: GLAMOR 0.5 Delivers Performance Improvements

    GLAMOR 0.5 was released earlier this month and testing has revealed that this 2D-Over-OpenGL acceleration library is now faster but still comes up short compared to other 2D driver implementations...

    http://www.phoronix.com/vr.php?view=MTE2Mjg

  2. #2
    Join Date
    Jul 2010
    Posts
    373

    Default

    Compared to Intel SNA, GLAMOR 0.5 would definitely get smacked, but there aren't results in the blog posting written by Chris Wilson.
    The second graph is relative to SNA.

  3. #3
    Join Date
    Dec 2011
    Posts
    1,932

    Default

    Hopefully in the future GLAMOR gets even faster now that it seems AMD have adopted it too.

    I wonder how much potential GLAMOR have if more people put more optimization efforts into it.

  4. #4
    Join Date
    Nov 2008
    Location
    Madison, WI, USA
    Posts
    838

    Default

    Looks like a pretty nice improvement. I have a feeling that a certain website owner/operator will be running comparisons between GLAMOR .5/.4 and the 2D-accel built into the radeon driver.... hopefully. ;-)

  5. #5
    Join Date
    Nov 2007
    Posts
    1,024

    Default

    I wonder how important glamour really is to the future. Should we expect good acceleration of the old X11 APIs, or should we expect apps and toolkits to directly drive GL with high performance rendering techniques?

    There's a lot of techniques and cases to concern oneself with when trying to get very high performance 2D rendering, especially as it comes to minimizing state changes and draw calls. I'm not sure how easily you can take an API designed around immediate mode rendering of primitives and images and transform it into batched and atlased rendering. Especially when it comes to the translucent effects so popular in modern 2D GUIs.

    Maybe it's very easy. I know in my field we just "do it right" from the start, but then we don't pretend to support legacy code in any way.

    Just a thought.

  6. #6
    Join Date
    Jul 2011
    Location
    florida usa
    Posts
    79

    Default

    glamor seems pointless to me. why the open source ati driver puts any efort into it is beyond me. we have gallium, and gallium has state trackers. if you want to accelerate 2d you just make a 2d gallium state tracker. like a opengv tracker. id make a cairo to opengv backend for cairo and i would rely on the xorg state tracker to do the few things that dont support cairo. if exa is too poor an api to make an efficiant state tracker with, just extend it like intel did with uxa.

    yes, i understrand that intel with their i965 driver doesnt use the gallium architechture and therefor all this is usless to htem, and they are using glamor as a bandaid to circumvent poor 2d performance, but, that doesnt mean we alll have to do that.

    i understand that maybe making a openvg backend for cairo isnt efficient because of differences between the 2 api's, or mesas openvg implementation sucks or whatever i may not have a lot of insight into. its just it sounds like a good way to do it since im pretty sure cairo is generaly a higher level api than openvg (im pretty shure openvg doesnt have nearly the font capabilities of cairo). maybe the correct way of doing it would just to be to make a cairo state tracker. i know some early phone devices didnt suport opengl but did suport openvg, so for such type of hardware it would allow 2d acceleration while causing least conflict.

  7. #7
    Join Date
    Sep 2009
    Posts
    114

    Default

    Any way to try Glamor with Nouveau? For some reason my 2D accel is pretty slow but 3D(for desktop composition) is fine.

  8. #8
    Join Date
    Dec 2011
    Posts
    1,932

    Default

    Quote Originally Posted by dalingrin View Post
    Any way to try Glamor with Nouveau? For some reason my 2D accel is pretty slow but 3D(for desktop composition) is fine.
    I don't think so. Not until GLAMOR is integrated with Nouveau. I don't think it is and have not heard any plans of so.

  9. #9
    Join Date
    Dec 2007
    Posts
    2,273

    Default

    Quote Originally Posted by benjamin545 View Post
    glamor seems pointless to me. why the open source ati driver puts any efort into it is beyond me. we have gallium, and gallium has state trackers. if you want to accelerate 2d you just make a 2d gallium state tracker. like a opengv tracker. id make a cairo to opengv backend for cairo and i would rely on the xorg state tracker to do the few things that dont support cairo. if exa is too poor an api to make an efficiant state tracker with, just extend it like intel did with uxa.
    Glamor already exists and accelerates a lot of things. Extending EXA or writing a new state tracker has yet to be done so that would need to be done first before you could even begin to optimize the stack or driver.

Posting Permissions

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