Announcement

Collapse
No announcement yet.

Did Hell Just Freeze Over? Here's Evergreen On Gallium3D!

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

  • #41
    Originally posted by bridgman View Post
    Is there any mechanism today that could be hacked to give a rough measure of GPU activity, say split between "busy", "flushing", and "idle" ? One of our suspicions was that the command submission/synchronization mechanism was not getting anywhere near as much pipelining between CPU and GPU activity as the proprietary drivers (although the proprietary drivers have it easier because they are multithreaded).
    Intel have a pretty nifty tool called intel-gpu-top which is pretty handy for showing what the GPU is busy doing. Couldn't something similar be developed for radeons?

    Comment


    • #42
      Definitely. There are a bunch of status bits showing which blocks are busy as well, so you can dig deeper into the chip as well. I'm just looking for something simple (and quick to implement) as a first step.
      Test signature

      Comment


      • #43
        Stinkin' edit limit.
        Test signature

        Comment


        • #44
          Sometimes it feels like half the messages here are at least partially about cursing the edit limit.

          Comment


          • #45
            The edit windows should be 5min. That enough to modify your post and keep the trolls out.

            Comment


            • #46
              Originally posted by bridgman View Post
              Is there any mechanism today that could be hacked to give a rough measure of GPU activity, say split between "busy", "flushing", and "idle" ?
              Nothing I know of.

              Originally posted by bridgman View Post
              There used to be a general consensus that redundant setting of state in each DRI2 command buffer (no lock so you don't know if the values you set last time are still valid) accounted for a non-trivial chunk of performance, is that still believed or has the theory been debunked ? Generating the state packets may not take a lot of CPU time but executing them on the GPU could take longer. If it is still an issue, we could probably do some context save/restore tricks (or just move context management into the kernel entirely) to speed things up.
              I think Dave has done some work to optimize redundant state setting at the beginning of CS and the performance difference was negligible, so I think this has been debunked.

              Comment


              • #47
                Originally posted by V!NCENT View Post
                The edit windows should be 5min. That enough to modify your post and keep the trolls out.
                Sorry, how exactly is edit window supposed to affect whether trolls get in or not? o.O

                Comment


                • #48
                  Just speaking for myself and my needs, r300g is very much fast enough. It's also coming very, very close to dethrone i965 as the free driver where most games/apps/demos work.

                  Comment


                  • #49
                    Originally posted by whizse View Post
                    Just speaking for myself and my needs, r300g is very much fast enough. It's also coming very, very close to dethrone i965 as the free driver where most games/apps/demos work.
                    Glad to hear that. Thanks.

                    Comment


                    • #50
                      Originally posted by nanonyme View Post
                      Sorry, how exactly is edit window supposed to affect whether trolls get in or not? o.O
                      My understanding was roughly the following :

                      New users aren't allowed to post link-laden ads (and other users don't seem to feel the need to do so ), but spammers were using the edit mechanism to make an initial post without links then go back and edit the post to add the rest of the advertisement. Limiting the edit window seemed to stop the practice.
                      Test signature

                      Comment

                      Working...
                      X