Announcement

Collapse
No announcement yet.

Mesa Lands Work Around Async glFlush - Should Help Workstation Performance

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

  • Mesa Lands Work Around Async glFlush - Should Help Workstation Performance

    Phoronix: Mesa Lands Work Around Async glFlush - Should Help Workstation Performance

    While vendors are increasingly just focused on Wayland (and Direct3D under Windows), there still is plenty of OpenGL software out there especially in the workstation space where software vendors are slow to adapt. Well known RadeonSI OpenGL developer Marek Olšák of AMD landed another performance optimization this week that should benefit the likes of SPECviewperf...

    https://www.phoronix.com/scan.php?pa...-Viewperf-Mesa

  • #2
    Nvidia had this for years. Famous kwin tearing on nvidia was caused by non-blocking glFlush on nvidia and mesa-specific glx usage in kwin. Gladly major compositors fixed issues regarding to async glFlush.
    Last edited by d3coder; 20 June 2021, 06:58 AM.

    Comment


    • #3
      > While vendors are increasingly just focused on Wayland

      I believe that was meant to say Vulkan, not Wayland?

      Comment


      • #4
        Originally posted by Black_Fox View Post
        > While vendors are increasingly just focused on Wayland

        I believe that was meant to say Vulkan, not Wayland?
        Yep hadn't enough coffee at that point, thanks.
        Michael Larabel
        http://www.michaellarabel.com/

        Comment


        • #5
          Is there any good reason why it wouldn't be in their interest to focus on Vulkan and OpenGL through Zink in the mid and long term? Like, do performance optimizations in a standalone OpenGL driver still makes sense, compared to investing this working time into Zink?

          Comment


          • #6
            Originally posted by Cr0ydon View Post
            Is there any good reason why it wouldn't be in their interest to focus on Vulkan and OpenGL through Zink in the mid and long term? Like, do performance optimizations in a standalone OpenGL driver still makes sense, compared to investing this working time into Zink?
            I suspect AMD is trying to dump their proprietary OGL driver on all platforms in favor of the open driver to reduce maintenance costs and improve driver quality all around. Zink still has quite a ways to go for competitive performance, whereas RadeonSI only really needs to fix workstation performance to be superior to the proprietary driver in every way.

            Comment


            • #7
              It seems unlikely to me that AMD would be able to abandon their proprietary OGL driver due to legacy customers who would have only validated their software on that driver stack. But at the same time it sounds like some new customers wanted AMD to invest in RadeonSI on Gallium3D, so maybe the customer base is split now.

              What I don't understand is why doesn't AMD bring RadeonSI/Gallium3D to windows for gaming purposes? That way AMD could have the best of both worlds. A gaming driver that's cross platform between Linux and Windows, while also having a proprietary stack that could be put in maintenance mode for legacy customers. People have been complaining about the performance of OpenGL on the Windows stack and it seems to me that this could be a potential solution for AMD.

              Comment


              • #8
                Originally posted by Cr0ydon View Post
                Is there any good reason why it wouldn't be in their interest to focus on Vulkan and OpenGL through Zink in the mid and long term? Like, do performance optimizations in a standalone OpenGL driver still makes sense, compared to investing this working time into Zink?
                They're talking about still needing to double performance of radeonsi in these old GL 1.1 apps in order to justify dumping their proprietary drivers. I'm guessing zink isn't anywhere close to what they need for their clients, and it may not even be possible. A shift to Zink is more likely to signal the complete abandonment of caring about OpenGL at all, and they aren't there yet.

                Comment

                Working...
                X