Announcement

Collapse
No announcement yet.

Experimental Code Gets Open-Source Mesa RADV Vulkan Driver Running On Windows

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

  • Experimental Code Gets Open-Source Mesa RADV Vulkan Driver Running On Windows

    Phoronix: Experimental Code Gets Open-Source Mesa RADV Vulkan Driver Running On Windows

    Over the years there have been various attempts at getting the open-source RADV Vulkan driver on Windows, Faith Ekstrand of Collabora has been recently hacking on it and achieving success for having this popular Radeon Vulkan API driver for Linux working under Windows...

    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
    This may be a stupid question, but why would you want to do this? Just because you can, or is there some practical benefit?

    Comment


    • #3
      Originally posted by J.King View Post
      This may be a stupid question, but why would you want to do this? Just because you can, or is there some practical benefit?
      RADV is generally a much better driver than AMDVLK on Linux, I see no reason why the same wouldn't be true on Windows as well.

      This could set an interesting precedent long-term, where OpenGL and Vulkan are provided by Mesa on all major OSes, not just Linux. We already are kinda seeing this with translation layers, e.g. with OpenGL on Windows, where Windows-on-Arm devices use the d3d12 Gallium driver in Mesa to provide OpenGL and OpenCL on Windows, so the manufacturer only has to provide a d3d12 driver and they get OpenGL and OpenCL for free.

      Comment


      • #4
        Originally posted by J.King View Post
        Just because you can?
        Indeed. It's the same useless task as getting RT to work on Polaris.

        No developer will adapt their games and software to an unofficial driver. That's not how it works in big business. The same thing happened with the Google Stadia, when to the surprise of many Linux fans, they used AMDVLK.

        Comment


        • #5
          Originally posted by J.King View Post
          This may be a stupid question, but why would you want to do this? Just because you can, or is there some practical benefit?
          AMD's Windows UMD are known to be of poor quality, and RADV may help with this. There is also hope for real graphics hardware acceleration (even ray tracing, mesh shader, cooperative matrix support) with RADV in WSL2 instead of crap like Dozen layered on D3D12.

          Comment


          • #6
            Originally posted by mphuZ View Post

            Indeed. It's the same useless task as getting RT to work on Polaris.

            No developer will adapt their games and software to an unofficial driver. That's not how it works in big business. The same thing happened with the Google Stadia, when to the surprise of many Linux fans, they used AMDVLK.
            It's the official driver of the Steam Deck, for what it's worth. Also, AMDVLK vs RADV was a much closer race when Stadia was just getting started.

            Comment


            • #7
              Also, think of game emulators that have a Vulkan backend, if they could just ship a copy of Mesa and get more performance and fewer bugs, why wouldn't they want to do that?

              Comment


              • #8
                Originally posted by QwertyChouskie View Post

                It's the official driver of the Steam Deck, for what it's worth.
                1. Today, developers practically do not create native ports for Linux, so it makes no sense to compare.
                2. Windows game developers hardly understand how RADV works themselves (corny due to the lack of Linux specialists). There is some verification from Valve that checks how Proton games work. I'm not sure if the developers are making any contribution to "porting" to Steam Deck/Proton or commits to the RADV driver.

                Comment


                • #9
                  I'd like to read a blog post about it, wanting to know the why and how and where it is heading. Is this a "for science" project or are there some specific goals behind it? Better performance and less bugs are always welcome if it seriously aims to de-throne AMD's official OpenGL and Vulkan driver on Windows in the long run.

                  Comment


                  • #10
                    I wonder if this could also lead towards improved VM guest support... there was something similar related to AMD GPUs with VM guests redirecting to the host GPU driver, but I think it was Linux only ... maybe with this progress it could also enable Windows support?

                    Comment

                    Working...
                    X