Announcement

Collapse
No announcement yet.

How The RadeonSI OpenGL Performance Has Evolved From Mesa 11.1 To Mesa 17.1 Git

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

  • How The RadeonSI OpenGL Performance Has Evolved From Mesa 11.1 To Mesa 17.1 Git

    Phoronix: How The RadeonSI OpenGL Performance Has Evolved From Mesa 11.1 To Mesa 17.1 Git

    For those curious how AMD's RadeonSI Gallium3D driver for GCN GPUs has evolved, here are benchmarks with two graphics cards showing how the RadeonSI Mesa performance has evolved since Mesa 11.1 going back to late 2015.

    http://www.phoronix.com/vr.php?view=24391

  • #2
    AMD's Fine Wine at its finest, although it probably applies to all Mesa drivers

    Comment


    • #3
      Originally posted by andrei_me View Post
      AMD's Fine Wine at its finest, although it probably applies to all Mesa drivers
      Fine Wine and Fine Line should be story for blobs. On opensource drivers it more like Fine Development actually, that starts from in some cases near unusable to more and more usable

      For both, if you buy new hardware but slighlty old gen at the end of the cycle things are Fine Immediately and we are free to complain why something else is still missing which is normal to have just because blobs have it, particulary in comparison to Windows
      Last edited by dungeon; 03-20-2017, 07:29 PM.

      Comment


      • #4
        I wonder how many little/annoying bugs were fixed between those versions besides perf improvements...

        Comment


        • #5
          Originally posted by andrei_me View Post
          I wonder how many little/annoying bugs were fixed between those versions besides perf improvements...
          Read mesa logs from last 10 months and llvm logs from the last about 16 months... that would clarify things.

          I don't think these games even render proper on that llvm 3.7... but never mind

          Comment


          • #6
            Originally posted by dungeon View Post
            For both, if you buy new hardware but slighlty old gen at the end of the cycle things are Fine Immediately
            it was true few cycles ago. now you don't need to buy old gen. maybe don't buy on first day and wait for reviews

            Comment


            • #7
              It is a shame that the Radeon R9 290 / 290X / 390 / 390X continue with the performance regression, regression started with the 4.7 kernel. Someone knows something about this problem??

              Comment


              • #8
                Originally posted by obedlink View Post
                Someone knows something about this problem??
                AFAIR, it was Ubuntu related, I remember some users here commenting that this issue didn't exists in Fedora or other distros.
                Maybe something related to the PPA used too, I don't remember if the out-of-box version of Ubuntu is regressed too...

                Comment


                • #9
                  Originally posted by obedlink View Post
                  It is a shame that the Radeon R9 290 / 290X / 390 / 390X continue with the performance regression, regression started with the 4.7 kernel. Someone knows something about this problem??
                  It is because of simultant two drivers for these, as devs does not care too much about radeon path but more about amdgpu for CI... should be general guess.

                  Just guessing since i think Michael reported R9 290 issue for radeon driver only, but amdgpu driver does not have that issue, etc...

                  It is simple bug, since it is regression bug someone who can reproduce it - should bisect it
                  Last edited by dungeon; 03-20-2017, 08:59 PM.

                  Comment


                  • #10
                    Originally posted by dungeon View Post

                    It is because of simultant two drivers for these, as devs does not care too much about radeon path but more about amdgpu for CI... should be general guess.

                    Just guessing since i think Michael reported R9 290 issue for radeon driver only, but amdgpu driver does not have that issue, etc...

                    It is simple bug, since it is regression bug someone who can reproduce it - should bisect it
                    If you were on mesa-devel someone has - has, again.

                    Seems to be BIOS/firmware thing, again:
                    bugzilla-daemon at freedesktop.org bugzilla-daemon at freedesktop.org [Mesa-dev] [Bug 97260] R9 290 low performance in Linux 4.7

                    https://lists.freedesktop.org/archiv...ch/148594.html

                    [-]
                    --- Comment #66 from tarpoon at gmx.de --- Okay. I bisected the issue today already: 7050c6ef5f0e9bc5e6bf9eb035320b70f731b919 is the first bad commit commit 7050c6ef5f0e9bc5e6bf9eb035320b70f731b919 Author: Arindam Nath <arindam.nath at amd.com> Date: Wed Apr 6 15:33:51 2016 -0400 drm/radeon: add support for loading new UVD fw Signed-off-by: Christian K├Ânig <christian.koenig at amd.com> Signed-off-by: Arindam Nath <arindam.nath at amd.com> Reviewed-by: Leo Liu <leo.liu at amd.com> Signed-off-by: Alex Deucher <alexander.deucher at amd.com> :040000 040000 25ff8fc8784746ea4e371a13655581a3e844e7b9 9016d0a57c0409e91eff182bbccd91510a8b85c1 M drivers [-] Later on: https://lists.freedesktop.org/archiv...ch/148753.html [-]
                    Shadow of Mordor Benchmark FPS: Kernel Version | Menu | AVG | MAX | MIN Kernel 4.6.7 | 100 | 58 | 161 | 30 Kernel 4.10 (with new Firmware) | 100 | 58 | 114 | 30 [-] Cheers! Dieter

                    Comment

                    Working...
                    X