Announcement

Collapse
No announcement yet.

Radeon KMS Gets Faster X-Video Support

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

  • Radeon KMS Gets Faster X-Video Support

    Phoronix: Radeon KMS Gets Faster X-Video Support

    If using the latest ATI open-source driver Git code for kernel mode-setting, there may be slightly better X-Video support without any visual slowdowns/problems. Pauli Nieminen has changed around the code so that a video buffer is allocated in the GTT (Graphics Translation Table) to provide faster memory operations that haven't been accelerated by KMS. The patch that provides the performance boost to X-Video under ATI kernel mode-setting can be found on the xorg-driver-ati mailing list and has since been committed...

    http://www.phoronix.com/vr.php?view=Nzk4NA

  • #2
    What they really need to speed up is basic 2D accel. Still, any speed improvement with KMS is a good thing.

    Comment


    • #3
      Originally posted by Melcar View Post
      What they really need to speed up is basic 2D accel. Still, any speed improvement with KMS is a good thing.
      EXA is already relatively fast, and even my lowly RV710 runs Compiz smoothly.

      Comment


      • #4
        With KMS? My HD4850 is chocking with simple 2D tasks. On my older card it's not as bad though.

        Comment


        • #5
          Xv was already working very good here. One of the areas where the oss driver really shines compared to fglrx. But of course any improvement is still welcome

          And I agree that EXA under KMS could use some speed improvement. I'm now in the situation where I have 2 screens, one fully accelerated EXA/Xv/3D, and one with only shadowfb 'acceleration'. Many 2D operations are much faster with shadowfb than with EXA acceleration.

          There are also at least two bug reports regarding this:

          http://bugzilla.kernel.org/show_bug.cgi?id=15293
          http://bugs.freedesktop.org/show_bug.cgi?id=26225

          Comment


          • #6
            Xv works fine for the most part, except for some videos where it simply won't play or cause the players to crash. That Planet Earth birds clip that everyone likes for example. Interestingly enough, if I use the mplayer build from here it plays fine; smooth playback actually, despite my CPU usage going up the roof (not even even fglrx plays that clip that well without having to resort to XvBA).
            And yeah, that FireFox issue has been driving me nuts. Good thing I have been moving to Chrome.

            Comment


            • #7
              Originally posted by Melcar View Post
              With KMS?
              No, I'm still running UMS until I get a legitimate reason to run KMS.

              Comment


              • #8
                Now we're waiting for some tests UMS vs KMS on Radeon cards R300-R700.

                Comment


                • #9
                  I did a small 2D test a while ago with my 200M. Been meaning to do one with my HD4850, but the damn thing takes so long to finish . Will cook up one for Xv as well latter this week.

                  Comment


                  • #10
                    Thanks Melcar. Unfortunately it doesn't look too optimistic for KMS (yet). I thought that now the performance of KMS with R300-R500 will be at least as good as UMS/non-KMS.

                    Comment


                    • #11
                      Originally posted by Melcar View Post
                      I did a small 2D test a while ago with my 200M. Been meaning to do one with my HD4850, but the damn thing takes so long to finish . Will cook up one for Xv as well latter this week.
                      Thanks for the figures. Hmm, doesn't look that bright with KMS enabled. Well, but then it's considered still being in development. So, I guess more waiting until the situation becomes better from the user's side will do.

                      Comment


                      • #12
                        I switched from fglrx drivers to radeon to get KMS some time ago, and got noticeable improvements in 2D (mainly scrolling and Kwin desktop effects) and video playback, as well as improved stability and fewer rendering bugs. I never did any actual benchmarks though, so this observation is more anecdotal than scientific.

                        Anyway, even if KMS isn't as good as UMS yet, it still beats fglrx, so it's really nothing to complain about...

                        Comment


                        • #13
                          Originally posted by DanL View Post
                          No, I'm still running UMS until I get a legitimate reason to run KMS.
                          I can give you two, at least for my workflow.
                          1) No delay or flickering when changeing virtual terminals (Ctrl+Alt+Fx)
                          2) OpenGL acceleration on more than one virtual terminal simultaneously!

                          Comment


                          • #14
                            Originally posted by DanL View Post
                            No, I'm still running UMS until I get a legitimate reason to run KMS.
                            The following isn't enough of a reason?

                            http://i45.tinypic.com/2udxphc.png

                            (OpenGL is not getting composited.)

                            Comment


                            • #15
                              DRI2 doesn't need KMS.

                              Comment

                              Working...
                              X