Announcement

Collapse
No announcement yet.

Dota 2 7.00 Benchmarks - Intel Vulkan vs. OpenGL On Linux - Mesa 13.1 + Linux 4.9

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

  • Dota 2 7.00 Benchmarks - Intel Vulkan vs. OpenGL On Linux - Mesa 13.1 + Linux 4.9

    Phoronix: Dota 2 7.00 Benchmarks - Intel Vulkan vs. OpenGL On Linux - Mesa 13.1 + Linux 4.9

    In addition to big end-of-year AMD Radeon Linux benchmarks and the forthcoming NVIDIA data points among other interesting EOY comparisons, there is also ongoing fresh Intel Linux benchmarks as we end out 2016. For your viewing pleasure today are the latest Intel OpenGL vs. Vulkan Linux benchmark results using last week's Dota 2 7.00 game release...

    http://www.phoronix.com/scan.php?pag...-Intel-VLK-OGL

  • #2
    Thanks for running these benchmarks, Michael!

    We've got some improvements coming which I hope will close this gap.

    Comment


    • #3
      Originally posted by Kayden View Post
      Thanks for running these benchmarks, Michael!

      We've got some improvements coming which I hope will close this gap.
      thats good to hear, specially when at 800x600 its already faster

      Comment


      • #4
        I know this benchmark is focused on Vulkan vs OpenGL, but this is how I feel when I see under 60fps at 800x600.

        Warning foul language...

        Comment


        • #5
          Michael regarding those test requests: what about selecting a few graphic cards (for example, RX 460/480 and a comparable NVidia card) and see how they run a few games/unigine tests on various CPU-motherboard-memory combinations? I am curious to see at which point the CPU/memory starts becoming the bottle neck for a given graphics card and game.

          Comment


          • #6
            Originally posted by Jabberwocky View Post
            I know this benchmark is focused on Vulkan vs OpenGL, but this is how I feel when I see under 60fps at 800x600.
            Squeeze down some game setting a bit and you will see more fps

            Uber maxing game settings out is generally reserved for fastest or sometimes even not yet released hardware

            Comment


            • #7
              Originally posted by Davidovitch View Post
              Michael regarding those test requests: what about selecting a few graphic cards (for example, RX 460/480 and a comparable NVidia card) and see how they run a few games/unigine tests on various CPU-motherboard-memory combinations? I am curious to see at which point the CPU/memory starts becoming the bottle neck for a given graphics card and game.
              He has already done that a couple of days ago with the 460, 470, 480 and Fury with RadeonSI.

              http://www.phoronix.com/scan.php?pag...ta2-700-radeon

              If you dig further back a few months I would be pretty sure that you would find a comparison between Intel, AMD and NVidia on a older Dota 2 version, but it shouldn't be more than a slight difference from the 7.00 so that is still be very relevant.

              Memory and motherboard shouldn't make a difference, but comparing with a slower CPU would be interesting. Hell, just locking down a few of those 8 cores would be interesting. A 2,3,4,6 and 8 core comparison with the same processor would be interesting to see.
              Last edited by johanb; 12-19-2016, 08:26 AM. Reason: grammar

              Comment


              • #8
                johanb do you mean this article (your link doesn't work): http://www.phoronix.com/scan.php?pag...0-radeon&num=1 ? Sure, but that one uses only one CPU: an Intel Xeon E3, and that CPU has been used in a lot of graphic card tests lately. I think it is a very good strategy to often tests various graphic cards with as fast a CPU available in order to avoid running into the limitations of the CPU instead of the GPU. However, once in a while it would be interesting to see how GPU performance varies as function of various less powerful CPU's. When buying a CPU+GPU you always have to compromise (due to limited available funds) between the GPU and the CPU, and I have no idea where such a sweet spot may be.
                Last edited by Davidovitch; 12-19-2016, 08:52 AM.

                Comment


                • #9
                  Originally posted by Jabberwocky View Post
                  I know this benchmark is focused on Vulkan vs OpenGL, but this is how I feel when I see under 60fps at 800x600.
                  I think the purpose of doing 800x600 is to skew the scales, so the CPU becomes more stressed than the GPU. One of the primary benefits of Vulkan is to reduce CPU overhead, so clearly at these lower resolutions, that seems to be working. It's too bad that it seems to be performing worse at higher resolutions though.

                  Comment

                  Working...
                  X