Announcement

Collapse
No announcement yet.

XRender on ati catalyst

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #21
    with fglrx 10.8 on 2.6.35

    firefox: 10
    chromium: 26

    Comment


    • #22
      Originally posted by bridgman View Post
      Sorry, just to be clear we have debug/verbose mechanisms for the developers, I thought you were asking for a way to see for yourself which functions were accelerated.

      Sounds like you are suggesting a verbose debug mechanism for regular users... other than "it works on our hardware but breaks on yours" scenarios how do you see that being useful ? Doesn't it need a developer to analyze the results anyways ?
      well is more like a tool to automatically report issues including heavy debug info (like kde crash system), cuz report in ati support "my games look funky when they use shaders" wont help anyone

      so having a tool like ./atisupport /path/mygame/binarie file --gui that internally recopilate the hardware used and run the apps in debug mode and maybe ask the user to take some screenshots, pack all that data and generate a report to the ati bug system

      in the case of xrender is doable too i think, ofc this case probably would be more on user demand thing.

      if i dont remember wrong there is a tool for the nvidia blob to do something like that in nvnews

      Comment


      • #23
        Originally posted by jrch2k8 View Post
        well is more like a tool to automatically report issues including heavy debug info (like kde crash system), cuz report in ati support "my games look funky when they use shaders" wont help anyone

        so having a tool like ./atisupport /path/mygame/binarie file --gui that internally recopilate the hardware used and run the apps in debug mode and maybe ask the user to take some screenshots, pack all that data and generate a report to the ati bug system

        in the case of xrender is doable too i think, ofc this case probably would be more on user demand thing.

        if i dont remember wrong there is a tool for the nvidia blob to do something like that in nvnews
        exactly !

        I used to report bugs with that in the former times when I still was having an nvidia card

        it dumps infos on dmesg, Xorg.0.log, nvidia drivers-version, afaik also the currently loaded modules and something more

        why not also provide such a tool - that sure would speed up the issue/bug-hunting

        Comment


        • #24
          Doing a dump on a crash is very different from dumping enough information to tackle performance optimization on a user system. One is fairly straightforward, the other is usually anything but.
          Test signature

          Comment


          • #25
            Originally posted by bridgman View Post
            Doing a dump on a crash is very different from dumping enough information to tackle performance optimization on a user system. One is fairly straightforward, the other is usually anything but.
            hm - ok - that shows my ignorance

            anyways this would be nice to have in the future to improve bug-hunting efficiency in general - wouldn't you think ?

            Comment


            • #26
              Originally posted by bridgman View Post
              Doing a dump on a crash is very different from dumping enough information to tackle performance optimization on a user system. One is fairly straightforward, the other is usually anything but.
              well i think you are partially rigth, from a dump is really hard to get something like your gfx is 3.2% slower in blit operation compared to X gpu we have here to develop, so i agree here.

              on the other hand is not necesarilly true that such diff will generate a massive slow down, the slowdowns are probably cuz a fallback to the cpu is happening for some X or Y reason and that can be traced with a bug report.

              so even if this tool is not for benchies, is helpful to get the most information needed to ironout the traceable bugs that probably helps to make thing go slower or like in opengl "funky shaders" without have to interpret what the users meant to report or to wait for some guy with the patience and lack of social life to debug the issues

              Comment


              • #27
                OK, I think you are saying that output from the driver trace could be used by an *application* developer to optimize their code, is that correct ?
                Test signature

                Comment


                • #28
                  Originally posted by bridgman View Post
                  OK, I think you are saying that output from the driver trace could be used by an *application* developer to optimize their code, is that correct ?
                  yes and i think is good for AMD devs of fglrx too, having a good tool to get the best debug possible could help you guys a lot to find faster bugs/fallbacks/etc + for application developers would be peachy to report issues too having that debug info

                  Comment


                  • #29
                    Again, we have debug tools for internal use... the discussion here is about the creation of a "sanitized" debug mechanism for use outside the AMD development environment.
                    Test signature

                    Comment


                    • #30
                      Originally posted by bridgman View Post
                      Again, we have debug tools for internal use... the discussion here is about the creation of a "sanitized" debug mechanism for use outside the AMD development environment.
                      yes exactly

                      Comment

                      Working...
                      X