Announcement

Collapse
No announcement yet.

Updated and Optimized Ubuntu Free Graphics Drivers

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

  • Update

    Here are some updates:
    • upgraded llvm to 3.6, currently at -rc1;
    • updated vaapi library and intel driver;
    • enabled support for shader cache (still not available upstream) adding SHA1 library.

    Still TODO updating libvdpau to 0.9.

    Comment


    • Originally posted by oibaf View Post
      Here are some updates:
      • upgraded llvm to 3.6, currently at -rc1;
      • updated vaapi library and intel driver;
      • enabled support for shader cache (still not available upstream) adding SHA1 library.

      Still TODO updating libvdpau to 0.9.
      Thank you very much for your work. I wish I could donate to you, but for now I can't...
      Anyway, I have a couple of questions: do you know if the fix for this bug has been submitted to mesa 10.5? https://bugs.freedesktop.org/show_bug.cgi?id=86837 , and if it hasn't, could you please apply that patch (https://bugs.freedesktop.org/show_bug.cgi?id=86837#c12) to your repo? cgit.freedesktop.org doesn't show anything related to this from Emil Velikov: http://cgit.freedesktop.org/mesa/mes...%40gmail%2ecom.

      Comment


      • Originally posted by asdfblah View Post
        Thank you very much for your work. I wish I could donate to you, but for now I can't...
        Anyway, I have a couple of questions: do you know if the fix for this bug has been submitted to mesa 10.5? https://bugs.freedesktop.org/show_bug.cgi?id=86837 , and if it hasn't, could you please apply that patch (https://bugs.freedesktop.org/show_bug.cgi?id=86837#c12) to your repo? cgit.freedesktop.org doesn't show anything related to this from Emil Velikov: http://cgit.freedesktop.org/mesa/mes...%40gmail%2ecom.
        I prefer avoiding adding patches that make the packages diverging from upstream unless really necessary. Best you should ask on the bug about the status of the patch.

        Comment


        • Nah that one still is not in master , that is 2 months vdpau being broken and just because libs wanted to be somewhat smaller

          radeon users should be more active on bugzilla and complain about that
          Last edited by dungeon; 01-24-2015, 03:50 AM.

          Comment


          • Originally posted by dungeon View Post
            Nah that one still is not in master , that is 2 months vdpau being broken and just because libs wanted to be somewhat smaller

            radeon users should be more active on bugzilla and complain about that
            You mean the ABI-compatibility thing where they don't want to put stuff in two unrelated places because there's no ABI checking anywhere so if they end up out of sync, things break in interesting ways? Tbh sounds like VDPAU should just be nuked in favour of the alternatives if the core design is that poor
            Last edited by nanonyme; 01-24-2015, 04:27 AM.

            Comment


            • Originally posted by nanonyme View Post
              You mean the ABI-compatibility thing where they don't want to put stuff in two unrelated places because there's no ABI checking anywhere so if they end up out of sync, things break in interesting ways? Tbh sounds like VDPAU should just be nuked in favour of the alternatives if the core design is that poor
              No i am thinking of, breaking whatever people use without working alternative in place is a no go... vaapi is also broken with this, not just vdpau .

              Comment


              • Besides both vaapi and classic vdpau (at least with mpv) works worse then vdpau with NV_vdpau_interop.
                Last edited by dungeon; 01-24-2015, 05:57 AM.

                Comment


                • @oibaf

                  After libvdpau update vdpauinfo. That matches usally all vdpau changes.

                  Comment


                  • Originally posted by oibaf View Post
                    I prefer avoiding adding patches that make the packages diverging from upstream unless really necessary. Best you should ask on the bug about the status of the patch.
                    OK, thank you.

                    Comment


                    • Originally posted by oibaf View Post
                      I prefer avoiding adding patches that make the packages diverging from upstream unless really necessary. Best you should ask on the bug about the status of the patch.
                      Isn't a broken vdpau+vaapi necessary enough?
                      ## VGA ##
                      AMD: X1950XTX, HD3870, HD5870
                      Intel: GMA45, HD3000 (Core i5 2500K)

                      Comment


                      • Hi! I've noticed LLVM 3.6 finally here. Which is great news for AMD GPUs users like me.

                        However it looks like if you've did something wrong. After perfrorming upgrade I'm getting errors like this:
                        cannot open file /usr/lib/clc/[name-of-my-GPU].bc - no such file or directory.

                        It appears that now package libclc-r600 from your repo only contains .bc files for old VLIW based GPUs but totally lacks any .bc files for GCN-based GPUs. As far as I understand now I have all code generation for my GCN GPU screwed, right? Am I missing something? Or is it some bug?

                        Comment


                        • Originally posted by darkbasic View Post
                          Isn't a broken vdpau+vaapi necessary enough?
                          Since this is broken for everyone, not only for this PPA, it should be applied upstream so that everyone could benefit .

                          Comment


                          • Originally posted by SystemCrasher View Post
                            Hi! I've noticed LLVM 3.6 finally here. Which is great news for AMD GPUs users like me.

                            However it looks like if you've did something wrong. After perfrorming upgrade I'm getting errors like this:
                            cannot open file /usr/lib/clc/[name-of-my-GPU].bc - no such file or directory.

                            It appears that now package libclc-r600 from your repo only contains .bc files for old VLIW based GPUs but totally lacks any .bc files for GCN-based GPUs. As far as I understand now I have all code generation for my GCN GPU screwed, right? Am I missing something? Or is it some bug?
                            Thanks for reporting, it should be fixed now .

                            Comment


                            • Guys is someone experiencing low fps in Dota2 settings?

                              I'm building my own Mesa, ddx, xserver and drm but my config is based on oibaf repository, a little bit tuned because I only need radeonSI.

                              I can't figure out why in settings and menus it runs at about 10FPS while when I start a match frames jump to 30FPS (luckly).

                              I'm sure back in November I can run at about 60FPS in settings (but lower than now in game, so it's still an improvement).

                              Comment


                              • Originally posted by lorenzoz View Post
                                Guys is someone experiencing low fps in Dota2 settings?

                                I'm building my own Mesa, ddx, xserver and drm but my config is based on oibaf repository, a little bit tuned because I only need radeonSI.

                                I can't figure out why in settings and menus it runs at about 10FPS while when I start a match frames jump to 30FPS (luckly).

                                I'm sure back in November I can run at about 60FPS in settings (but lower than now in game, so it's still an improvement).
                                Are you sure the hardware acceleration is enabled in your system? Run this command, if you see "VMware" then you are running the software renderer:
                                Code:
                                $ glxinfo | grep render
                                If not, I think you should submit a bug report.
                                IIRC, Mesa 10.5 had a performance problem, but I can't remember more info...

                                Comment

                                Working...
                                X