Announcement

Collapse
No announcement yet.

NVIDIA 375 vs. RADV+RadeonSI Mesa 13.1-dev Vulkan Benchmarks For Ending 2016

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

  • NVIDIA 375 vs. RADV+RadeonSI Mesa 13.1-dev Vulkan Benchmarks For Ending 2016

    Phoronix: NVIDIA 375 vs. RADV+RadeonSI Mesa 13.1-dev Vulkan Benchmarks For Ending 2016

    The latest installment of our year-end benchmarks is focusing upon the performance of the NVIDIA Linux driver against the open-source Radeon Vulkan (RADV) driver found within Mesa 13.1-dev. This comparison is particularly interesting given the continuous flow of improvements into Mesa Git, the NVIDIA 375.26 driver release from last week, the big Dota 2 7.00 update debuted earlier this month, and Croteam's Vulkan improvements have rolled into TTP stable.

    http://www.phoronix.com/vr.php?view=23899

  • #2
    What sort of settings were you using on the Talos Principle to get the performance down that low? I regularly exceed 130 FPS using Vulkan at 2560x1440 resolution levels.

    Comment


    • #3
      Talos numbers look like 4K reather than 1080, can you verify ?

      Comment


      • #4
        Originally posted by nocri View Post
        Talos numbers look like 4K reather than 1080, can you verify ?
        That does sound more plausible.

        Comment


        • #5
          Talos perf is really bad. This is really weird.

          Comment


          • #6
            Michael the sentences are mixed up:
            AMDGPU DRM for RADV compatibility include the GeForce GTX 650, GTX 680, GTX 780 Ti, GTX 950, GTX 960, GTX 970, GTX 980, GTX 980 Ti, GTX 1060, GTX 1070, and GTX 1080.
            On the NVIDIA side was the 375.26 driver as their latest stable release as of one week ago. Tested cards there were the R9 285, RX 460, RX 480, and R9 Fury.

            More typos:

            while the RX 480 was running just ahead of a RX 480
            Should be GTX 960
            Last edited by andrei_me; 12-20-2016, 05:57 PM.

            Comment


            • #7
              Why GTX 650 was failing with Vulkan? No Dota and Talos benchmarks.

              Comment


              • #8
                I concur with other users with Talos Principle. I get 30-40 FPS average for 64-Bit OpenGL and this is not during the benchmark. The CPU preset is set to high and the rest is set to Ultra with 2x anti-aliasing and 3D Rendering MPIX set to 2.1 (1080P). When going through the elevator in any world areas, I get 60+ FPS and did hit 110+ FPS for about a second or two before the elevator stops at the top. I'm running nvidia-lqx ver. 375.26-1 in Linux Liquorix kernel 4.8.10-1, although I will upgrade soon to get the security fixes but then I have not fall victim to exploits from criminals.

                So yeah, it would be great if you show us the settings.
                Last edited by GraysonPeddie; 12-20-2016, 08:25 PM.

                Comment


                • #9
                  He, he, Michael must be playing with server currently... sometimes openbenchmarking go down sometimes entire phoronix

                  Originally posted by GraysonPeddie View Post
                  So yeah, it would be great if you show us the settings.
                  Config should be here, but currently OB is down:

                  https://openbenchmarking.org/test/pts/talos-principle

                  But it have startup autodetection enabled last time i saw it, so settings likely get changed at runtime

                  Also worth to mention is that Talos beta builds are slower then release builds, probably more debuging enabled... so if someone run that it is expected to be slower
                  Last edited by dungeon; 12-21-2016, 12:36 AM.

                  Comment


                  • #10
                    I don't see what's interesting about comparing the propriety driver of one manufacturer and opensource of the other. It's comparing apples with pears. I find it more interesting how the propriety driver compares to the opensource driver of the same manufacturer.

                    Comment

                    Working...
                    X