Announcement

Collapse
No announcement yet.

NVIDIA vs. AMD OpenGL & Vulkan Benchmarks With Valve's Dota 2

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

  • NVIDIA vs. AMD OpenGL & Vulkan Benchmarks With Valve's Dota 2

    Phoronix: NVIDIA vs. AMD OpenGL & Vulkan Benchmarks With Valve's Dota 2

    Yesterday marked the public availability of Dota 2 with a Vulkan renderer after Valve had been showing it off for months. This is the second commercial Linux game (after The Talos Principle) to sport a Vulkan renderer and thus we were quite excited to see how this Dota 2 Vulkan DLC is performing for both NVIDIA GeForce and AMD Radeon graphics cards. Here are our initial Dota 2 benchmarks with Vulkan as well as OpenGL for reference when using the latest Linux graphics drivers on Ubuntu.

    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
    Should have tested 800x600 not something that's most likely GPU bound already like 4K.

    Comment


    • #3
      Unlikely but, ... could it be that the game benefits from having more than 4 GB of video memory? That's what's different between the GTX 980 Ti and the other GPUs.

      Comment


      • #4
        Test are completely fine. Dota 2 is not graphical intensive and nor require more than 4g vram in 4k. As we know nvidia driver quality is better from amd also vulkan renderer of dota 2 is incomplete

        Comment


        • #5
          I'm laughing at the 980 getting spanked by R9 285.

          Comment


          • #6
            Still a long way to go for vulkan games.

            Comment


            • #7
              Originally posted by blackout23 View Post
              Should have tested 800x600 not something that's most likely GPU bound already like 4K.
              Why? Benchmarking in resolutions no one will ever use is a waste of time.

              Originally posted by boffo View Post
              Still a long way to go for vulkan games.
              Don't expect any miracles from game engines supporting multiple APIs sinche they use an abstraction layer. There will probably never be a single game with an optimal OpenGL and Vulkan pipeline, so benchmarking that is kind of pointless.

              Comment


              • #8
                Why are you benchmarking a new API whose primary purpose is to reduce CPU at 4K? All you are doing is measuring fill-rate and ensuring that the app is 100% GPU bound and CPU sitting mostly idle. If you actually want to see how Vulkan improves over OpenGL, you need to benchmark at much lower resolutions.
                Last edited by slacka; 24 May 2016, 05:32 PM.

                Comment


                • #9
                  Hm, disappointing that it is noticeably slower with Vulkan. With Talos Principle, the AMD GPUs were noticeably faster, but then, Dota 2 works very well with OpenGL on radeon, but not so much with the Talos Principle.

                  There are no AMD numbers due to the AMDGPU-PRO driver not currently exposing the GPU utilization in any meaningful way.
                  Isn't it using a slightly modified amdgpu kernel module and libdrm?
                  Wouldn't
                  Code:
                  [FONT=monospace][COLOR=#000000]radeontop -d -[/COLOR][/FONT]
                  or so work?

                  As for the 980ti discrepancy: It would be cool if the phoronix test suite could automatically produce a couple of screenshots while benchmarking, so the actual graphical output could be compared.

                  Comment


                  • #10
                    Originally posted by slacka View Post
                    Why are you benchmarking a new API whose primary purpose is to reduce CPU at 4K? All you are doing is measuring fill-rate and ensuring that the app is 100% GPU bound and CPU sitting mostly idle. If you actually want to see how Vulkan improves over OpenGL, you need to benchmark at much lower resolutions.
                    even my smartphone has an HD resolution. Makes no sense to test at lower resolution than HD. If this is what it was designed to do than it is an inferior design to opengl.

                    Comment

                    Working...
                    X