Announcement

Collapse
No announcement yet.

Performance Work Coming Up For Mesa 7.11

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

  • #21
    Originally posted by RealNC View Post
    Tell that to the "blobs must die, only open source drivers should be allowed in Linux" bozos.
    It's a nice thought though! If everybody just got along...

    Comment


    • #22
      I want to thanks the devs for their great effort!

      THANKS!!!

      Comment


      • #23
        Originally posted by hal2k1 View Post
        Gallium drivers are a work in progress. The comparisons that you linked to do not include any of the performance improvements that are the subject of this thread, nor do they include Page Flipping which will not be available until Linux kernel 2.6.38 is released.
        Not 100% sure about this, but I think Phoronix test suite sets vblank_mode to 0 and pageflipping is only used with vblank_mode=1, so there wouldn't probably be any visible change.

        Comment


        • #24
          Originally posted by RealNC View Post
          Tell that to the "blobs must die, only open source drivers should be allowed in Linux" bozos.
          There is no reason why open source drivers cannot be as fast as the closed blobs in principle. Opening the code does not make it slower, and neither does the license.

          It's just that the number of OSS driver developers is so low, that the devs have opted for a design which improves maintainability and code reuse over hardware-specific tuning.

          I can live with that. I can't live with the "kill all OSS drivers" psycho gamerz crowd.

          Comment


          • #25
            Exactly. From a Linux perspective, proprietary drivers have only one real advantage -- they can share development effort with proprietary drivers for other OSes and in that way benefit from a much larger development effort.

            Comment


            • #26
              Originally posted by RealNC View Post
              Tell that to the "blobs must die, only open source drivers should be allowed in Linux" bozos.
              Did it ever occur to you that those people gladly sacrifice speed for Opensource drivers?

              Comment


              • #27
                Originally posted by Paulie889 View Post
                Not 100% sure about this, but I think Phoronix test suite sets vblank_mode to 0 and pageflipping is only used with vblank_mode=1, so there wouldn't probably be any visible change.
                Not sure about this, but if that is the case then ordinary users will possibly see a significant performance gain in open source Radeon drivers with Linux kernel 2.6.38 where Phoronix tests show none.

                What good then are Phoronix tests?

                Comment


                • #28
                  Pageflipping as it is currently implemented in dri2 only flips during the vblank period (swap interval of 1) so it's limited to the refresh rate. Ideally, we also support pageflipping when the swap interval was 0 as well so you'd get atomic pageflipping (with tearing) for maximum frame rates. dri2 needs to work to enable this however. Even with vblank synchronized pageflipping, it still helps performance in that is reduces memory bandwidth requirements and pipeline latency since you don't have to do an additional blit to copy the back buffer to the front buffer.

                  Comment


                  • #29
                    Originally posted by glxextxexlg View Post
                    The big thanks goes to Marek Olsak and Corbin Simpson, who resurrected the r300-r500 hardware with their outstanding efforts and without the support of a giant company behind them.]
                    They definitely deserve our thanks for all the hard work, but AMD has supported them and others by providing docs and answers/clarifications on technical issues from our internal hw/sw teams.

                    Comment


                    • #30
                      Originally posted by RealNC View Post
                      Tell that to the "blobs must die, only open source drivers should be allowed in Linux" bozos.
                      There's not blobs must die only open source drivers should be allowed in linux bozos. What there are is an operating system controls your computer and takes responsibility for controlling your computer. You shouldn't have 3rd parties in there doing things you don't fully understand. You'll reverse enegineer a good deal of what they are doing just by working with them. But they are still an unknown ingredient in your sauce. And they cause testing problems because if something in their package causes a problem in your package then you have to search around in the dark for an answer and then wait for them to fix it.

                      I know redhat is very hardcore about security and I'm sure working with some of these companies gives you a whole lot of conciousness that you are working with pedo bear and want to start screaming stranger danger for some odd reason. All you have to do is watch things like the fence sync. They always operate the same way. Create a crisis and then solve it quickly not giving everyone enough time to figure out everything that is going on and why it's going on. They are addicted to creating crisis because they already know how they want it solved and they want it solved with everybody throwing thier hands up in disgust.

                      Comment

                      Working...
                      X