Announcement

Collapse
No announcement yet.

Updated and Optimized Ubuntu Free Graphics Drivers

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

  • Yes.. Seems the last intel update solve the problem..
    But DRI_PRIME/OFFLOAD is 100% broken.
    So way of load using xrandr

    Comment


    • I was testing out the mesa AMD OpenCL drivers with pyrit benchmarks, and I was surprised by the performance gains vs the proprietary AMD drivers: 20-30%! Gallium OpenCL has gotten a lot better in the past few months.

      So, I was wondering if the maintainer of the repository wouldn't mind applying this patch I made (adapted from this debian sid patch) to turn on opencl support in mesa? This patch builds out the opencl libraries to seperate deb packages, and it is tested and working on the latest mesa oibaf source package (as of right now). It can be found here: http://pastebin.com/iaaRCtBm

      The only other thing required is the libclc libraries, which can be backported from the trusty working repo here. The llvm and clang packages in the oibaf repository already fulfill the requirements.

      Regardless, thank you so much for this repo, it has saved me a lot of time and anguish in trying to get the latest mesa drivers working.

      Comment


      • Originally posted by fritzls View Post
        Yes.. Seems the last intel update solve the problem..
        But DRI_PRIME/OFFLOAD is 100% broken.
        So way of load using xrandr
        Well yeah, the issue with Chrome has been fixed, but another one (ones?) appeared. This happened on my laptop, and on workstation. After some time X crashes, and I get lightdm login screen appears. Unfortunately I don't have time to report the issue, because I cannot find where the logs are : ). It looks to me that log files structure has changed on Ubuntu. I have just checked again and this is all I have found:

        Jan 29 19:28:31 denis-desktop kernel: [ 3.355118] [drm] Wrong MCH_SSKPD value: 0x16040307
        Jan 29 19:28:31 denis-desktop kernel: [ 3.355120] [drm] This can cause pipe underruns and display issues.
        Jan 29 19:28:31 denis-desktop kernel: [ 3.355121] [drm] Please upgrade your BIOS to fix this.
        Jan 29 19:28:31 denis-desktop kernel: [ 3.700331] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device
        Jan 29 19:28:31 denis-desktop kernel: [ 3.700333] i915 0000:00:02.0: registered panic notifier
        Jan 29 19:28:31 denis-desktop kernel: [ 3.700348] i915: No ACPI video bus found
        Jan 29 19:28:31 denis-desktop kernel: [ 3.700375] [drm] Initialized i915 1.6.0 20080730 for 0000:00:02.0 on minor 0
        Jan 29 19:28:40 denis-desktop /usr/sbin/gpm[1999]: *** info [daemon/startup.c(131)]:
        Jan 29 19:28:40 denis-desktop /usr/sbin/gpm[1999]: Started gpm successfully. Entered daemon mode.

        Since I have some work to do, I have purged oibaf repo, and will try in couple of days again.

        Comment


        • Originally posted by reCAPTCHA View Post
          Well yeah, the issue with Chrome has been fixed, but another one (ones?) appeared. This happened on my laptop, and on workstation. After some time X crashes, and I get lightdm login screen appears. Unfortunately I don't have time to report the issue, because I cannot find where the logs are : ). It looks to me that log files structure has changed on Ubuntu. I have just checked again and this is all I have found:

          Jan 29 19:28:31 denis-desktop kernel: [ 3.355118] [drm] Wrong MCH_SSKPD value: 0x16040307
          Jan 29 19:28:31 denis-desktop kernel: [ 3.355120] [drm] This can cause pipe underruns and display issues.
          Jan 29 19:28:31 denis-desktop kernel: [ 3.355121] [drm] Please upgrade your BIOS to fix this.
          Jan 29 19:28:31 denis-desktop kernel: [ 3.700331] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device
          Jan 29 19:28:31 denis-desktop kernel: [ 3.700333] i915 0000:00:02.0: registered panic notifier
          Jan 29 19:28:31 denis-desktop kernel: [ 3.700348] i915: No ACPI video bus found
          Jan 29 19:28:31 denis-desktop kernel: [ 3.700375] [drm] Initialized i915 1.6.0 20080730 for 0000:00:02.0 on minor 0
          Jan 29 19:28:40 denis-desktop /usr/sbin/gpm[1999]: *** info [daemon/startup.c(131)]:
          Jan 29 19:28:40 denis-desktop /usr/sbin/gpm[1999]: Started gpm successfully. Entered daemon mode.

          Since I have some work to do, I have purged oibaf repo, and will try in couple of days again.
          I am blind. I see now exactly the same crash message as you had.

          Jan 29 21:52:26 denis-desktop kernel: [ 8642.843435] [drm] stuck on render ring
          Jan 29 21:52:26 denis-desktop kernel: [ 8642.843442] [drm] GPU crash dump saved to /sys/class/drm/card0/error
          Jan 29 21:52:26 denis-desktop kernel: [ 8642.843443] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
          Jan 29 21:52:26 denis-desktop kernel: [ 8642.843444] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
          Jan 29 21:52:26 denis-desktop kernel: [ 8642.843444] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
          Jan 29 21:52:26 denis-desktop kernel: [ 8642.843445] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.

          But the error file /sys/class/drm/card0/error has nothing in it... Only "no error state collected".

          Comment


          • Originally posted by Doctor Nick View Post
            I was testing out the mesa AMD OpenCL drivers with pyrit benchmarks, and I was surprised by the performance gains vs the proprietary AMD drivers: 20-30%! Gallium OpenCL has gotten a lot better in the past few months.

            So, I was wondering if the maintainer of the repository wouldn't mind applying this patch I made (adapted from this debian sid patch) to turn on opencl support in mesa? This patch builds out the opencl libraries to seperate deb packages, and it is tested and working on the latest mesa oibaf source package (as of right now). It can be found here: http://pastebin.com/iaaRCtBm

            The only other thing required is the libclc libraries, which can be backported from the trusty working repo here. The llvm and clang packages in the oibaf repository already fulfill the requirements.

            Regardless, thank you so much for this repo, it has saved me a lot of time and anguish in trying to get the latest mesa drivers working.
            I was aware of the debian report but I never tried it, now I added it to the PPA (based on the debian bug patch, your, the only notable difference I see was using clang 3.4, didn't apply cleanly). Thanks for the interest anyway!

            The package to use is libopencl1-mesa. Let me/us know if/how it works!

            Comment


            • Originally posted by oibaf View Post
              I was aware of the debian report but I never tried it, now I added it to the PPA (based on the debian bug patch, your, the only notable difference I see was using clang 3.4, didn't apply cleanly). Thanks for the interest anyway!

              The package to use is libopencl1-mesa. Let me/us know if/how it works!
              Woop! Yes, I just installed it and ran the libclc and pyrit selftests, everything seems to be working fine. Thank you!

              Yeah, I had to edit that patch by hand because it seems that I'm retarded when it comes to editing and pushing changes to debian source packages . Do you know any good resources so I can figure out what the hell I'm doing?

              BTW, if any of you want to run pyrit using mesa-opencl, you have to build cpyrit from source and edit setup.py to include /usr/include/CL, since it doesn't look in there when searching for the opencl headers. Build BOTH pyrit and cpyrit-opencl from source, if you install the pyrit package, it'll fuck up the library search path for cpyrit and it won't load.

              Comment


              • Last updates seems solve this and dri_prime/offload.

                The onl thing that bug me is I cant use YET the offload to load the ubuntu by default or browsers.
                Ubuntu load exporting globally the dri_prime but no background ( black ones) and sites inside the browser ( flash or not) freezes.

                Wish one day use my 6850m inside notebook to do something more than NOTHING ( I'm not a gamer).


                Originally posted by reCAPTCHA View Post
                I am blind. I see now exactly the same crash message as you had.

                Jan 29 21:52:26 denis-desktop kernel: [ 8642.843435] [drm] stuck on render ring
                Jan 29 21:52:26 denis-desktop kernel: [ 8642.843442] [drm] GPU crash dump saved to /sys/class/drm/card0/error
                Jan 29 21:52:26 denis-desktop kernel: [ 8642.843443] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
                Jan 29 21:52:26 denis-desktop kernel: [ 8642.843444] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
                Jan 29 21:52:26 denis-desktop kernel: [ 8642.843444] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
                Jan 29 21:52:26 denis-desktop kernel: [ 8642.843445] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.

                But the error file /sys/class/drm/card0/error has nothing in it... Only "no error state collected".

                Comment


                • Originally posted by oibaf View Post
                  I was aware of the debian report but I never tried it, now I added it to the PPA (based on the debian bug patch, your, the only notable difference I see was using clang 3.4, didn't apply cleanly). Thanks for the interest anyway!

                  The package to use is libopencl1-mesa. Let me/us know if/how it works!
                  wow, thank you very much!

                  Comment


                  • good morning

                    my notebook :
                    core-i5 M 480
                    AMD Radeon HD 6550M
                    8 Gb Ram
                    Xubuntu 13.10 X86_64 + Oibaf ppa

                    I upgrade to kernel 3.13.1 and now I can not make the switch
                    sudo cat /sys/kernel/debug/vgaswitcheroo/switch
                    Code:
                    0:IGD:+:Pwr:0000:00:02.0
                    1:DIS: :DynOff:0000:01:00.0
                    2:DIS-Audio: :Off:0000:01:00.1
                    before upgrade :
                    Code:
                    0:IGD:+:Pwr:0000:00:02.0
                    1:DIS: :Pwr:0000:01:00.0
                    2:DIS-Audio: :Pwr:0000:01:00.1

                    echo ON > /sys/kernel/debug/vgaswitcheroo/switch
                    does not work


                    echo DDIS > /sys/kernel/debug/vgaswitcheroo/switch
                    restart lightdm

                    lightdm does not restart
                    I read this message: client refuse switch

                    is this a bug?

                    Comment


                    • Originally posted by Far03 View Post
                      good morning

                      my notebook :
                      core-i5 M 480
                      AMD Radeon HD 6550M
                      8 Gb Ram
                      Xubuntu 13.10 X86_64 + Oibaf ppa

                      I upgrade to kernel 3.13.1 and now I can not make the switch
                      sudo cat /sys/kernel/debug/vgaswitcheroo/switch
                      Code:
                      0:IGD:+:Pwr:0000:00:02.0
                      1:DIS: :DynOff:0000:01:00.0
                      2:DIS-Audio: :Off:0000:01:00.1
                      before upgrade :
                      Code:
                      0:IGD:+:Pwr:0000:00:02.0
                      1:DIS: :Pwr:0000:01:00.0
                      2:DIS-Audio: :Pwr:0000:01:00.1

                      echo ON > /sys/kernel/debug/vgaswitcheroo/switch
                      does not work


                      echo DDIS > /sys/kernel/debug/vgaswitcheroo/switch
                      restart lightdm

                      lightdm does not restart
                      I read this message: client refuse switch

                      is this a bug?
                      Switching is done automatically on 3.13 and newer. The dGPU is automatically powered down when not in use. You can get the old manual behaviour back by booting with radeon.runpm=0 on the kernel command line in grub.

                      Comment

                      Working...
                      X