Announcement

Collapse
No announcement yet.

Intel X.Org Driver Now Handles Better Tear-Free

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

  • #11
    Originally posted by Bucic View Post
    And what is this tearfree you speak of? It doesn't seem to be a package or a tool.
    It is an Option for the driver:
    Code:
    Section "Device"
      Identifier "Device0"
      Option "TearFree" "true"
    EndSection
    Originally posted by sharan View Post
    when will opengl 3.3 and these feature get in xf86-video-ati
    -ati already offers something similar: Option "VSync". However, that causes the driver to stall the GPU prior to every operation that writes to the frontbuffer - potentially causing a severe performance degradation - and only working correctly for the primary output. As for OpenGL 3.3, your crystal ball is as good as mine.

    Comment


    • #12
      I'd love for the annoying 'font glitch', where a random letter appears all garbled to go away on the gm 45 since it's pretty much the only i?sue I came across using SNA, not sure it will happen tho since it's old hardware....


      I ask the intel devs not to forget, or neglect, iron lake, gen4 and gma 950 .... I still use them daily

      Comment


      • #13
        Originally posted by ickle View Post
        As for OpenGL 3.3, your crystal ball is as good as mine.
        Ickle, two questions for ya if you don't mind.

        1) Any word, that you can speak of, from the higher ups about whether or not Intel is going to adopt Nvidia's new G-Sync tech?

        2) As someone who hasn't worked with OpenGL an awful lot, I know that getting the higher OpenGL versions supported is important for features, but I was curious... Going from 3.0 to 3.3 or 3.3 to 4.0 or any other OpenGL version bump, do they actually yield performance improvements? 2.1 to 3.0 I'm sure had a lot of improvements just because of how much of a break it was. But from 3.0 forward has Khronos ever released an extension that was basically "This is the same as the old one...but we found a better way to do it, so its faster." or a similar situation wherein just by supporting a higher OpenGL spec the performance would increase.
        All opinions are my own not those of my employer if you know who they are.

        Comment


        • #14
          Dear ickle,
          I know it's enerving to get and answer all the questions about bugs but are there any updates regarding: https://bugs.freedesktop.org/show_bug.cgi?id=54226 ?

          In the meantime i will give the new driver release a try

          Comment


          • #15
            Thank you, Shirudo and Ickle!

            Originally posted by Shirudo View Post
            It's a configuration option to remove screen tearing. From the Arch forums:

            cat /etc/X11/xorg.conf.d/20-intel.conf
            Section "Device"
            Identifier "Intel Graphics"
            Option "SwapbuffersWait" "true"
            Option "AccelMethod" "sna"
            Option "TearFree" "true"
            EndSection

            I've used these options myself and they work perfectly, I think I get better performance from switching to sna too.
            Excuse my ognorance but why, the fuck, TearFree is not enabled by default?! Does someone up there think tear-free image is a luxury in 2013 Linux?

            I may well sound like a radical, but to me work on a module that introduced a regression causing tearing should be halted and all resources assigned to the task until the regression is removed. That would teach those interns. Sorry, but if the word 'tearing' comes up in 2013 this alone indicates a failure. And here we have Linus Torvalds ranting over Fedora not refreshing installation images. Adjust your freaking priorities!
            Last edited by Bucic; 23 October 2013, 02:04 PM.

            Comment


            • #16
              Originally posted by Pallidus View Post
              I'd love for the annoying 'font glitch', where a random letter appears all garbled to go away on the gm 45 since it's pretty much the only i?sue I came across using SNA, not sure it will happen tho since it's old hardware....


              I ask the intel devs not to forget, or neglect, iron lake, gen4 and gma 950 .... I still use them daily
              HOLY CRAP. THIS IS CAUSE BY SNA?? I've been freaking out over what has been causing this forever! I'm using Manjaro and when I set TearFree to true and enabled sna (like the Arch thingy told me to) I started having these issues. I was certain it was caused by FreeType2 being shit... but it happened with FreeType2-Ifinality and FreeType2-Ubuntu as well Q__Q

              Excuse my ognorance but why, the fuck, TearFree is not enabled by default?! Does someone up there think tear-free image is a luxury in 2013 Linux?

              I may well sound like a radical, but to me work on a module that introduced a regression causing tearing should be halted and all resources assigned to the task until the regression is removed. That would teach those interns. Sorry, but if the word 'tearing' comes up in 2013 this alone indicates a failure. And here we have Linus Torvalds ranting over Fedora not refreshing installation images. Adjust your freaking priorities!
              I use Windows 7 sometimes, and I'd just like you to know that an up-to-date Win7 install has screen tearing... worse than on my Linux install when I don't have TearFree set to true!! And no tearing at all when it's enabled :P

              Comment


              • #17
                Originally posted by Daktyl198 View Post
                HOLY CRAP. THIS IS CAUSE BY SNA?? I've been freaking out over what has been causing this forever! I'm using Manjaro and when I set TearFree to true and enabled sna (like the Arch thingy told me to) I started having these issues. I was certain it was caused by FreeType2 being shit... but it happened with FreeType2-Ifinality and FreeType2-Ubuntu as well Q__Q
                To be fair, it is a GPU bug... Except that they officially do not exist. (UXA triggers the same bugs as well, except there I used a very big hammer and spent the time developing SNA to try to characterise the problem and find a better solution.)

                Originally posted by theghost View Post
                I know it's enerving to get and answer all the questions about bugs but are there any updates regarding: https://bugs.freedesktop.org/show_bug.cgi?id=54226 ?
                I know nothing new. Every suggestion I have made turned out to have no effect and I haven't found anything else in the bspec we could try. We can't turn off semaphores as that just reintroduces a different hardware bug (which again do not exist). We are stuck between a rock and a hard place - I just hope that one day we see a pattern in the hangs which would hopefully lead us to a solution. Until then, I am very sorry.

                Originally posted by Ericg View Post
                1) Any word, that you can speak of, from the higher ups about whether or not Intel is going to adopt Nvidia's new G-Sync tech?
                I haven't heard anything yet, but the technique does look quite familiar - it looks pretty similar to the PSR that is employed to reduce power consumption, except here they are using it to increase performance and eliminate output latency. Very cool idea. Anyway, I have no idea yet how Intel will respond - I just hope it is quickly!

                Originally posted by Ericg View Post
                2) As someone who hasn't worked with OpenGL an awful lot, I know that getting the higher OpenGL versions supported is important for features, but I was curious... Going from 3.0 to 3.3 or 3.3 to 4.0 or any other OpenGL version bump, do they actually yield performance improvements? 2.1 to 3.0 I'm sure had a lot of improvements just because of how much of a break it was. But from 3.0 forward has Khronos ever released an extension that was basically "This is the same as the old one...but we found a better way to do it, so its faster." or a similar situation wherein just by supporting a higher OpenGL spec the performance would increase.
                Since even 3.0, there has been a dramatic change in the nature of the GPU - mostly due to the ever increasing number of specialised programmable stages in the pipeline, as well as increasing their versatility (e.g. OpenCL). The biggest changes in each recent OpenGL version have basically been hardware enablement. Though there is the usual refinement in the shader languages, API and functionality as well, most of the improvement is to make new techniques possible.

                Comment


                • #18
                  Sadly it appears that 2.99.905 broke the driver on my System 76 gazp9 laptop with a Haswell Intel 4600. I reverted back to 2.99.904 and the driver works again. I was building new versions directly out of git regularly without issues. My last pre 905 build was done two to three days ago. So some change committed in the meantime totally broke the driver for me ;(

                  EDIT: For what it's worth I believe my last build occurred at this particular commit:

                  Last edited by jaylittle; 23 October 2013, 07:50 PM.

                  Comment


                  • #19
                    Crap it must've been an earlier revision as I just tried it with that one and I still have the issue <sigh>

                    EDIT: The issue seems to have something to do with the custom modelines I am using. 2.99.904 detects them whereas 2.99.905 does not. My initial res is set to something less than the native resolution of the LCD (1920x1080 max versus 1600x900 set). I wonder if this is the issue at hand...
                    Last edited by jaylittle; 23 October 2013, 08:01 PM.

                    Comment


                    • #20
                      Here is a comparison of my two XOrg log files. The first one is for 2.99.904 and the second one is 2.99.905:

                      Code:
                      [  4641.378] (II) intel(0): EDID vendor "LGD", prod id 933
                      [  4641.378] (II) intel(0): Printing DDC gathered Modelines:
                      [  4641.378] (II) intel(0): Modeline "1920x1080"x0.0  144.20  1920 1968 2040 2148  1080 1083 1088 1120 -hsync -vsync (67.1 kHz eP)
                      [  4641.652] (II) intel(0): EDID vendor "LGD", prod id 933
                      [  4641.652] (II) intel(0): Printing DDC gathered Modelines:
                      [  4641.652] (II) intel(0): Modeline "1920x1080"x0.0  144.20  1920 1968 2040 2148  1080 1083 1088 1120 -hsync -vsync (67.1 kHz eP)
                      [  4648.653] (II) intel(0): EDID vendor "LGD", prod id 933
                      [  4648.653] (II) intel(0): Printing DDC gathered Modelines:
                      [  4648.653] (II) intel(0): Modeline "1920x1080"x0.0  144.20  1920 1968 2040 2148  1080 1083 1088 1120 -hsync -vsync (67.1 kHz eP)
                      [  4649.676] (II) intel(0): resizing framebuffer to 1600x900
                      [  4649.676] (II) intel(0): switch to mode [email protected] on pipe 0 using eDP1, position (0, 0), rotation normal
                      [  4650.342] (II) intel(0): EDID vendor "LGD", prod id 933
                      [  4650.342] (II) intel(0): Printing DDC gathered Modelines:
                      [  4650.342] (II) intel(0): Modeline "1920x1080"x0.0  144.20  1920 1968 2040 2148  1080 1083 1088 1120 -hsync -vsync (67.1 kHz eP)
                      [  4650.556] (II) intel(0): EDID vendor "LGD", prod id 933
                      [  4650.556] (II) intel(0): Printing DDC gathered Modelines:
                      [  4650.556] (II) intel(0): Modeline "1920x1080"x0.0  144.20  1920 1968 2040 2148  1080 1083 1088 1120 -hsync -vsync (67.1 kHz eP)
                      [  4650.623] (II) intel(0): EDID vendor "LGD", prod id 933
                      [  4650.623] (II) intel(0): Printing DDC gathered Modelines:
                      [  4650.623] (II) intel(0): Modeline "1920x1080"x0.0  144.20  1920 1968 2040 2148  1080 1083 1088 1120 -hsync -vsync (67.1 kHz eP)
                      [  4651.581] (II) intel(0): EDID vendor "LGD", prod id 933
                      [  4651.581] (II) intel(0): Printing DDC gathered Modelines:
                      [  4651.582] (II) intel(0): Modeline "1920x1080"x0.0  144.20  1920 1968 2040 2148  1080 1083 1088 1120 -hsync -vsync (67.1 kHz eP)
                      [  4671.307] (II) intel(0): EDID vendor "LGD", prod id 933
                      [  4671.307] (II) intel(0): Printing DDC gathered Modelines:
                      [  4671.307] (II) intel(0): Modeline "1920x1080"x0.0  144.20  1920 1968 2040 2148  1080 1083 1088 1120 -hsync -vsync (67.1 kHz eP)
                      Code:
                      [  4612.109] (II) intel(0): EDID vendor "LGD", prod id 933
                      [  4612.109] (II) intel(0): Printing DDC gathered Modelines:
                      [  4612.109] (II) intel(0): Modeline "1920x1080"x0.0  144.20  1920 1968 2040 2148  1080 1083 1088 1120 -hsync -vsync (67.1 kHz eP)
                      [  4619.123] (II) AIGLX: Suspending AIGLX clients for VT switch
                      [  4640.846] (II) UnloadModule: "synaptics"
                      [  4640.846] (II) evdev: AT Translated Set 2 keyboard: Close
                      [  4640.846] (II) UnloadModule: "evdev"
                      [  4640.846] (II) evdev: BisonCam, NB Pro: Close
                      [  4640.846] (II) UnloadModule: "evdev"
                      [  4640.846] (II) evdev: Sleep Button: Close
                      [  4640.846] (II) UnloadModule: "evdev"
                      [  4640.846] (II) evdev: Power Button: Close
                      [  4640.846] (II) UnloadModule: "evdev"
                      [  4640.846] (II) evdev: Video Bus: Close
                      [  4640.846] (II) UnloadModule: "evdev"
                      [  4640.846] (II) evdev: Power Button: Close
                      [  4640.846] (II) UnloadModule: "evdev"
                      [  4640.857] (EE) Server terminated successfully (0). Closing log file.
                      Before this point, everything is the same.

                      Comment

                      Working...
                      X