Announcement

Collapse
No announcement yet.

Zink With Mesa 21.1 Now Advertises OpenGL 4.6

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

  • Zink With Mesa 21.1 Now Advertises OpenGL 4.6

    Phoronix: Zink With Mesa 21.1 Now Advertises OpenGL 4.6

    Mike Blumenkrantz continues to be on a mad roll when it comes to getting all of the Zink patches upstreamed into mainline Mesa... This Gallium3D-based OpenGL over Vulkan translation layer now has OpenGL 4.6 turned on for Mesa 21.1!..

    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
    Wow, that's impressive. A blink of an eye and it's on 4.6 already.

    Comment


    • #3
      Originally posted by tildearrow View Post
      Wow, that's impressive. A blink of an eye and it's on 4.6 already.
      Because it's only about advertising certain support but not actually a working implementation. This seems to be motivated by the developer's recent decision to start testing Zink using real PC games, which in turn seem to require advertising latest OpenGL feature levels.
      Last edited by curfew; 16 February 2021, 04:16 AM.

      Comment


      • #4
        Impressive!

        I'd like to hear a bit more about this only being an artificial version bump? curfew
        As on mesamatrix it also looks like all extensions are really implemented!

        Regardless, still darn impressive!

        I'm curious if they eventually get OpenGL working at "native" speeds through Vulkan. And if they do, what value does the native implementation still has?

        Usually a layer of indirection (vulkan is that layer in this case) puts you a little further away from the actual hardware you're talking too and therefore adds overhead (and thus reduced speed).

        Comment


        • #5
          It is still at OpenGL ES 3.1 though, so I hope to see OpenGL ES 3.2 soon. Also there are also some extensions outside of the standard that could be implemented.

          Great to see it at OpenGL 4.6, I hope to see performance improvements now! 👍
          Show Mesa progress for the OpenGL implementation into an easy to read HTML page.

          Comment


          • #6
            I guess the idea is that most games and software written for OpenGL with no Vulcan renderer were written with the performance of previous generation GPUs in mind, or the current ones at best. Going forward, faster future GPUs will eat up the performance hit of the Zink layer, making it a compatibility layer for old (when looking back from the future) software that even with the performance hit of Zink will run far faster than they were designed for. Eventually, there will be no more native OpenGL drivers.

            Comment


            • #7
              Originally posted by Tuxie View Post
              I guess the idea is that most games and software written for OpenGL with no Vulcan renderer were written with the performance of previous generation GPUs in mind, or the current ones at best. Going forward, faster future GPUs will eat up the performance hit of the Zink layer, making it a compatibility layer for old (when looking back from the future) software that even with the performance hit of Zink will run far faster than they were designed for. Eventually, there will be no more native OpenGL drivers.
              I will only add that CPUs will have to eat up that performance hit as well as overhead is not only on GPU side when it comes to OpenGL to Vulkan translation. Perhaps it's even mostly on CPU side.
              Last edited by reavertm; 16 February 2021, 07:36 AM.

              Comment


              • #8
                Originally posted by Tuxie View Post
                I guess the idea is that most games and software written for OpenGL with no Vulcan renderer were written with the performance of previous generation GPUs in mind, or the current ones at best. Going forward, faster future GPUs will eat up the performance hit of the Zink layer, making it a compatibility layer for old (when looking back from the future) software that even with the performance hit of Zink will run far faster than they were designed for. Eventually, there will be no more native OpenGL drivers.
                Its going to be a while before we can retire the existing Opengl drivers. Remember you still find Integrated Matrox G200eW Graphic in server motherboards as new so 20 year old design requiring basically drivers that support 20 year old hardware. So eventually no more opengl drivers is possible longer than your total life-time times 2.

                Its surprising to lots of people how many really old designs keep on turn up like a bad penny as there are cost reasons for them..

                Please don't think Matrox G200eW is the worst either at least it has something of a GPU.
                https://www.asrockind.com/en-gb/M2-VGA-IPC Yes old school vga upgraded for PCIe made for your M.2 slot with only 64MB of ram for video buffers and is basically GPU less so CPU software rendering required the chipset is still being made brand new and appearing on motherboards.

                Zink for opengl most likely only going to be useful for the graphics out solutions that can support Vulkan. The horrible hardware like the Matrox G200eW possible opengl drivers for those and of course for the I am bare min stuff software rendering opengl and vulkan will be required at times.

                Tuxie those making the really horrible video outs with absolute bare min required to make a signal is going to keep on doing so. So these horrible will keep on turning up because that the least amount of silicon solution so the least amount of cost. So I don't see opengl drivers going away completely but in time if you have a graphics out needing a opengl driver its going to be the in the low performing class built for cost not performance using some patent and royalty expired design. Yes these horrible are going to be your video out on servers and embedded devices for management reasons a interface you hope normal desktop users are not using in 20-100 years time but the servers and embedded serous-ally could be using 20-100 years into the future at least..

                Comment


                • #9
                  Originally posted by oiaohm View Post

                  Its going to be a while before we can retire the existing Opengl drivers. Remember you still find Integrated Matrox G200eW Graphic in server motherboards as new so 20 year old design requiring basically drivers that support 20 year old hardware. So eventually no more opengl drivers is possible longer than your total life-time times 2.

                  Its surprising to lots of people how many really old designs keep on turn up like a bad penny as there are cost reasons for them..

                  Please don't think Matrox G200eW is the worst either at least it has something of a GPU.
                  https://www.asrockind.com/en-gb/M2-VGA-IPC Yes old school vga upgraded for PCIe made for your M.2 slot with only 64MB of ram for video buffers and is basically GPU less so CPU software rendering required the chipset is still being made brand new and appearing on motherboards.

                  Zink for opengl most likely only going to be useful for the graphics out solutions that can support Vulkan. The horrible hardware like the Matrox G200eW possible opengl drivers for those and of course for the I am bare min stuff software rendering opengl and vulkan will be required at times.

                  Tuxie those making the really horrible video outs with absolute bare min required to make a signal is going to keep on doing so. So these horrible will keep on turning up because that the least amount of silicon solution so the least amount of cost. So I don't see opengl drivers going away completely but in time if you have a graphics out needing a opengl driver its going to be the in the low performing class built for cost not performance using some patent and royalty expired design. Yes these horrible are going to be your video out on servers and embedded devices for management reasons a interface you hope normal desktop users are not using in 20-100 years time but the servers and embedded serous-ally could be using 20-100 years into the future at least..
                  I think Codeweavers may have a screw loose somewhere. First they talking MAC but they use extensions not available on MAC. Then they talking old hardware but there is no MAC on this hardware neither the extensions they use. I think hiring a good management will be the best here.

                  Comment


                  • #10
                    I wonder how long it's gonna take for this to beat the Windows opengl implementation of AMD in term of performance. The contrast of their Vulkan implantation vs their OpenGL is so damn big that I can't believe it wont happen.

                    Comment

                    Working...
                    X