Announcement

Collapse
No announcement yet.

Direct3D 9 Support Proposed For DXVK

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

  • Direct3D 9 Support Proposed For DXVK

    Phoronix: Direct3D 9 Support Proposed For DXVK

    DXVK has been doing great with its Direct3D 11 support mapped to Vulkan and running a variety of games at performant speeds under Wine while now patches have emerged that would add Direct3D 9 support too...

    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
    I wonder why such stack of abstractions is considered. Every layer adds a possible source of bugs. I hope they have a really good reason for doing so.

    Comment


    • #3
      If it almost reaches the performance like DX11 via DXVK, to run DX9 via DX11 is fantastic news :-)

      Comment


      • #4
        I don’t have much knowledge about graphics apis and almost 0 about directX but isn’t DirectX 9, 11, 12 has common parts? It’s tough to think that instead of 3 separate projects to map them to vulkan a unified approach would help them in long run.

        Comment


        • #5
          Originally posted by bitman View Post
          I wonder why such stack of abstractions is considered. Every layer adds a possible source of bugs. I hope they have a really good reason for doing so.
          On the other hand, building upon a tried and tested layer that's doing the most complicated work sounds like a better idea than rebuilding everything from scratch, doesn't it? Both for the sake of development cost and avoiding bugs.

          Plus, they can test with DX11 on Windows.

          Originally posted by Tiger_Coder View Post
          I don’t have much knowledge about Igraphics apis and almost 0 about directX but isn’t DirectX 9, 11, 12 has common parts? It’s tough to think that instead of 3 separate projects to map them to vulkan a unified approach would help them in long run.

          More like D3D 9, 10 and 11 are somewhat iterative improvements. 12 is quite a big departure from the previous architecture, or so I'm told; which better maps to Vulkan/Mantle/Metal.
          DXUP maps D3D10 to D3D11, this would do the same for D3D9. This isn't really 3 separate projects, since DXVK ends up doing the bulk of the work, the other layers "just" make deprecated concepts and interfaces available to applications.
          Last edited by M@yeulC; 09 August 2018, 10:53 AM.

          Comment


          • #6
            Originally posted by Tiger_Coder View Post
            I don’t have much knowledge about graphics apis and almost 0 about directX but isn’t DirectX 9, 11, 12 has common parts?
            Probably, but there were some major differences going from d3d9 to d3d10 and then again with d3d12.

            Comment


            • #7
              Originally posted by M@yeulC View Post

              On the other hand, building upon a tried and tested layer that's doing the most complicated work sounds like a better idea than rebuilding everything from scratch, doesn't it? Both for the sake of development cost and avoiding bugs.

              Plus, they can test with DX11 on Windows.




              More like D3D 9, 10 and 11 are somewhat iterative improvements. 12 is quite a big departure from the previous architecture, or so I'm told; which better maps to Vulkan/Mantle/Metal.
              DXUP maps D3D10 to D3D11, this would do the same for D3D9. This isn't really 3 separate projects, since DXVK ends up doing the bulk of the work, the other layers "just" make deprecated concepts and interfaces available to applications.
              (Disclaimer: I am no expert myself in that area)

              IIRC DX9, 10 and 11 do share a more common heritage.

              IIRC DX12 lifted (stole) ideas from AMD's Mantle and is completely different than DX9-11. I thought they rebuilt it from the ground up.

              So AFAIK DX12 has certain similarities to Mantle (Which is now Vulkan) which is funny because IMO it excells in no particular way other than having an established API name.

              IIRC there aren't really all that many DX12 games right now so doing a DXVK12 is mostly pointless until there's exclusive content we want to play on Linux.

              Comment


              • #8
                I think dxup for dx10->dx11 is ok, but isn't dx9 a lot more different ? Also why not support vk9 or gallium-nine instead?

                Comment


                • #9
                  Originally posted by M@yeulC View Post

                  On the other hand, building upon a tried and tested layer that's doing the most complicated work sounds like a better idea than rebuilding everything from scratch, doesn't it? Both for the sake of development cost and avoiding bugs.

                  Plus, they can test with DX11 on Windows.




                  More like D3D 9, 10 and 11 are somewhat iterative improvements. 12 is quite a big departure from the previous architecture, or so I'm told; which better maps to Vulkan/Mantle/Metal.
                  DXUP maps D3D10 to D3D11, this would do the same for D3D9. This isn't really 3 separate projects, since DXVK ends up doing the bulk of the work, the other layers "just" make deprecated concepts and interfaces available to applications.
                  My understanding is DX9->DX10 was also a significant gap in terms on hardware capabilities. One of the problems with supporting that many generations is you bang hardware with vastly different feature sets and the possibility to share code reduces constantly.

                  Comment


                  • #10
                    I'll take anything that can make Guild Wars 2 (DX9) on Wine less of a performance mess.

                    Comment

                    Working...
                    X