Announcement

Collapse
No announcement yet.

Bettering Radeon Gallium3D Performance With PCI-E 2.0

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

  • Bettering Radeon Gallium3D Performance With PCI-E 2.0

    Phoronix: Bettering Radeon Gallium3D Performance With PCI-E 2.0

    While it will not take you up to the speeds of the Catalyst driver, besides the 2D color tiling patches, there are a few other outstanding features not yet enabled-by-default in the open-source Radeon graphics driver that can yield some performance boosts. One of these other features is enabling PCI Express 2.0 support within the Radeon DRM.

    http://www.phoronix.com/vr.php?view=16963

  • #2
    Surprised this hasn't been mentioned before. It seems like a fairly big win overall and for once, it's already ready for us to use! I suspect this will work for me so I'll try it later.

    Comment


    • #3
      i test it with my hd4770 works perfect for me.

      but just to be sure how to check the status of the pcie mode?

      Comment


      • #4
        Is there a correlation to the amount of VRAM?

        Comment


        • #5
          Just in time for PCI-E 3.0
          Didn't image such a big boost, at least in Windows the boost is very little with pci-e 2.0
          ## VGA ##
          AMD: X1950XTX, HD3870, HD5870
          Intel: GMA45, HD3000 (Core i5 2500K)

          Comment


          • #6
            This performance gain is amazing, I mean up by 54%! I guess the bandwidth is really only needed when pushing much data, as in higher resolutions and large textures.. Or am I wrong? I will enable this today if I get the chance! Thank you for yet another awesome tip Michael. This alone is worth my Phoronix Subscription =)

            Comment


            • #7
              Interesting tests, thank you Michael!
              This is exactly why I like to come to Phoronix!

              Comment


              • #8
                actually im a little dissappointed. not by the boost, but by the fact i didnt know about it before. i was like "how can the catalyst driver be 10x faster?" there was no word about missing pci-e spec features in that article.
                what id like to see now would be a comparison of catalyst with radeon, on single core cpus (with pci-e enabled to the highest of course) and all the hardware features enabled that would potentially boost hardware performance in the oss driver.
                what other features are there, ready or yet to be met?
                (like higher, not default gpu stages maybe?)
                maybe a list of missing features would be good...
                is there a performance gain to be expected when the driver someday gets capable of multithreading? if so what would be the gain like?
                could there be the answer to the cpu bottleneck that seems to appear?

                what is the actual difference between catalyst abd the open source driver? and dont tell me its only about optimisation...

                Comment


                • #9
                  Among others hyper-z and 2d tiling which are being worked on. They will provide a big boost but there are so many things missing that I will be old before reaching parity with catalyst.
                  ## VGA ##
                  AMD: X1950XTX, HD3870, HD5870
                  Intel: GMA45, HD3000 (Core i5 2500K)

                  Comment


                  • #10
                    Originally posted by darkbasic View Post
                    Among others hyper-z and 2d tiling which are being worked on. They will provide a big boost but there are so many things missing that I will be old before reaching parity with catalyst.
                    Unless they start merging stuff from Catalyst to the FOSS drivers. I don't quite understand why they don't do that. I assume that the FOSS devs have access to the closed source code for Catalyst. I mean, what's the difference between merging actual Catalyst code and re-implementing it from open specifications? In the end, the functionality (and to some extent the code) should be the same.

                    Comment

                    Working...
                    X