Announcement

Collapse
No announcement yet.

Nouveau X.Org Driver Sees First Release In Two Years

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

  • Nouveau X.Org Driver Sees First Release In Two Years

    Phoronix: Nouveau X.Org Driver Sees First Release In Two Years

    Two years and nine patches later, xf86-video-nouveau 1.0.17 is out as the latest X.Org driver update for this open-source NVIDIA driver component...

    Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite

  • #2
    Is there any reason to use these drivers instead of the generic modesetting one? (besides hardware-specific features, of course)

    Comment


    • #3
      With modesetting, the hardware cursor tends to stutter/block on some occasions, whereas that's not the case with xf86-video-amdgpu.
      Also, you don't need to compile the whole xserver when you want newer commits, which is now almost (or entirely?) equally dead in terms of new releases.

      Comment


      • #4
        Originally posted by tildearrow View Post
        Is there any reason to use these drivers instead of the generic modesetting one? (besides hardware-specific features, of course)
        I always overlook the modesetting drivers (when on amd64 anyway). Does the modesetting one provide any acceleration? If so, it would surely be using parts of the Nouveau driver.

        If I was just going to use libreoffice, I wonder which one would be a better choice?

        Comment


        • #5
          Always use xf86-video-nouveau, never -modesetting. Unfortunately some major distros (Fedora, Debian) have taken the inverse approach, which leads to considerable instability for nouveau users.

          The difference is that xf86-video-nouveau is simple, well-tested, and has solid error handling. -modesetting uses GLAMOR which is a full-on GL application, makes use of various advanced features, and (obviously) relies on the GL component. While we've tried to make the GL driver good, it's missing some key bits that make it inappropriate for a long-running application like Xorg.

          Comment


          • #6
            thanks, updated: https://t2sde.org/packages/xf86-video-nouveau

            Comment


            • #7
              Originally posted by imirkin View Post
              Always use xf86-video-nouveau, never -modesetting. Unfortunately some major distros (Fedora, Debian) have taken the inverse approach, which leads to considerable instability for nouveau users.

              The difference is that xf86-video-nouveau is simple, well-tested, and has solid error handling. -modesetting uses GLAMOR which is a full-on GL application, makes use of various advanced features, and (obviously) relies on the GL component. While we've tried to make the GL driver good, it's missing some key bits that make it inappropriate for a long-running application like Xorg.
              I'm a bit disappointed with this answer, while I understand that nouveau development requires a lot of effort (and even more so because of the RE aspect). I don't think we should tell users to use xf86-video-nouveau because of limitations in nouveau (when -modesetting could also be made to work).

              I personally use and prefer the -modesetting driver with amdgpu and Intel i915 and I would like to use it with nouveau as well.

              Comment


              • #8
                Originally posted by ihatemichael View Post

                I'm a bit disappointed with this answer, while I understand that nouveau development requires a lot of effort (and even more so because of the RE aspect). I don't think we should tell users to use xf86-video-nouveau because of limitations in nouveau (when -modesetting could also be made to work).

                I personally use and prefer the -modesetting driver with amdgpu and Intel i915 and I would like to use it with nouveau as well.
                Huh? To rephrase what you're saying "we shouldn't tell users what works and what doesn't, because the thing that doesn't *should* work, and it might in the future!" That doesn't seem right.

                Comment

                Working...
                X