Announcement

Collapse
No announcement yet.

AMDGPU-PRO Radeon RX 460/470/480 vs. NVIDIA Linux GPU Benchmarks

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

  • AMDGPU-PRO Radeon RX 460/470/480 vs. NVIDIA Linux GPU Benchmarks

    Phoronix: AMDGPU-PRO Radeon RX 460/470/480 vs. NVIDIA Linux GPU Benchmarks

    Last week I published an 18-way GPU Linux comparison featuring the new Radeon RX 460 and RX 470 graphics cards along with other AMD Radeon and NVIDIA GeForce GPUs. The Radeon tests were done using the very latest open-source Linux driver stack while in this article are similar benchmarks done but using the AMDGPU-PRO hybrid driver stack.

    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
    Pretty disappointing showing from AMD. With the exception of Xonotic, the GTX1060 beats out all of AMDs offerings. And here in Canada a GTX1060 and RX480 are the same price. I really want to love AMD for their great work on the open-source drivers, and was hoping AMD-PRO would close the gap in a pinch. But the GTX series is well and truly trouncing the RX series in these tests. Paying a little extra money, or losing 15% of the frame rate to get a card with a good opensource driver option is worth it in my opinion. But losing 41 to 69 FPS in Metro Last Light Redux on a card that is the same price? That's a hard pill to swallow.

    Comment


    • #3
      Finally, vulkan is looking pretty good for radeon. Nvidia cuts corners on specs/standards so a lower overhead for them makes sense. Curious how the new mesa git stacks up in OpenGL against the PRO driver. I'm already seeing a massive improvement in ~minimum~ fps for games. ~+45% fps in saints row 4 and a ~+15% improvement in shadow warrior.

      Comment


      • #4
        Nice testing. What about throwing in Tomb Raider & Shadow of Mordor?

        Looking forward to the day AMD drivers reach hardware-performance parity. When do you think that will happen? 5.0+?

        Comment


        • #5
          Originally posted by Xen0sys View Post
          Nice testing. What about throwing in Tomb Raider & Shadow of Mordor?
          They aren't fully automated thus not tested, only occasionally when enough Phoronix Premium members / tippers request such.
          Michael Larabel
          https://www.michaellarabel.com/

          Comment


          • #6
            Originally posted by Michael View Post

            They aren't fully automated thus not tested, only occasionally when enough Phoronix Premium members / tippers request such.
            Oh that's a shame since they both have integrated benchmarks. Could it be automated with a little keyboard input magic ? O.o

            Comment


            • #7
              Originally posted by Xen0sys View Post

              Oh that's a shame since they both have integrated benchmarks. Could it be automated with a little keyboard input magic ? O.o
              Haven't yet found a universal way to achieve that due to different resolutions, etc. Plus for forcing the correct visual settings, etc, via the CLI.
              Michael Larabel
              https://www.michaellarabel.com/

              Comment


              • #8
                Originally posted by Michael View Post

                Haven't yet found a universal way to achieve that due to different resolutions, etc. Plus for forcing the correct visual settings, etc, via the CLI.
                True that would be messy. Spit-balling, but maybe there's a config file somewhere that could be edited to take care of the settings. Then it could just come down to presetting how many down's & enter's to get it going.

                Comment


                • #9
                  Originally posted by Michael View Post

                  Haven't yet found a universal way to achieve that due to different resolutions, etc. Plus for forcing the correct visual settings, etc, via the CLI.
                  A virtual controller should solve that. Controller input in usually 100% independent of screen resolution, you don't have to click certain parts of the screen. A hack sure, but install dirs being hardcoded is an ugly hack too.

                  Comment


                  • #10
                    Great set of benchmark you did Michael. I just gave a tip for it. Thanks for all

                    Comment

                    Working...
                    X