Announcement

Collapse
No announcement yet.

12-Way Linux Graphics Card Comparison Using The Newest May 2018 Drivers

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

  • 12-Way Linux Graphics Card Comparison Using The Newest May 2018 Drivers

    Phoronix: 12-Way Linux Graphics Card Comparison Using The Newest May 2018 Drivers

    Here is a look at twelve different AMD Radeon and NVIDIA GeForce graphics cards while testing was done using the newest available graphics drivers and using an Ubuntu 18.04 LTS installation.

    http://www.phoronix.com/vr.php?view=26343

  • #2
    Thanks Michael. If it not too much inconvenience would be great for AMDGPU-PRO Vulkan results next to RADV. But that's just a nitpick.
    The 580 looks like best value for the money. And while in Windows it's just a tiny below 1060, on Linux it crushes the 1060 both in OpenGl and Vulkan.
    Waiting now for Vega to shine also. Congratulation once again to AMD open source driver developers.

    Unrelated: Congratulation to Real Madrid for the 13th Champions Cup win.

    Comment


    • #3
      Strange, I thought the 290 was a lot closer in performance to the 580. I know the 580 is pretty much always faster, but not by this much.

      Comment


      • #4
        I haved rx 580 monthS ago, with cs:go cpu game around 250 fps (maybe with last mesa its better now), and r9 290 140-185. You can see 30-40% more fps for rx 580.


        offtopic:
        No see cs:go in benchmark for long time.
        I know cs:go is cpu game, its just for compare amd card vs nvidia cards with cs:go cpu game.

        I have r9 290, around 140-185 in 1920x1080 or 2560x1440 12 threads +-4 ghz, with last mesa kernel etc.

        I appreciate to see if gtx 1050 ti can perform better than my r9 290, but benchmark is always with gtx 1060.
        Someone have gtx 1050 ti ? what result ?
        I dont want buy gtx 1050 ti if same fps r9 290, i play only cs, cs:go.
        And interest gtx 970 secondhand same price new gtx 1050 ti.
        Last edited by kripteks; 05-26-2018, 09:49 PM.

        Comment


        • #5
          Originally posted by valici View Post
          The 580 looks like best value for the money. And while in Windows it's just a tiny below 1060, on Linux it crushes the 1060 both in OpenGl and Vulkan. Waiting now for Vega to shine also. Congratulation once again to AMD open source driver developers.
          Absolutely agree! I did a quick calculation using the data here. Leaving out those results with insanely high fps, as well as anything with unplayable low fps, I found that the RX 580 is on average 10% faster than the 1060, 41% faster than the 290 and 6% slower than the Fury.

          It's mighty impressive, especially when comparing it with the Fury. If only I can get my hands on one for a reasonable price!

          Comment


          • #6
            Great benchmarks thanks. If possible, please include Counter-strike: Global Offensive in the next run and ideally show the minimum fps values for it.

            Comment


            • #7
              Originally posted by arunbupathy View Post

              Absolutely agree! I did a quick calculation using the data here. Leaving out those results with insanely high fps, as well as anything with unplayable low fps, I found that the RX 580 is on average 10% faster than the 1060, 41% faster than the 290 and 6% slower than the Fury.

              It's mighty impressive, especially when comparing it with the Fury. If only I can get my hands on one for a reasonable price!
              In my country 1060 6Gb is 20% cheaper than RX580, so how it can be best ?

              Comment


              • #8
                Originally posted by cybertraveler View Post
                Great benchmarks thanks. If possible, please include Counter-strike: Global Offensive in the next run and ideally show the minimum fps values for it.
                There is a long-standing bug in CS:GO (or one could also argue GNOME behavior) yielding issues with automated benchmarking when triggered under GNOME Shell thinks the app is locked up. It's long been known and seems to only affect GNOME and is due to some net wm class handling or something if my memory serves me... but is still unresolved so that's why the Ubuntu switch to GNOME that I haven't been running CS:GO tests.
                Michael Larabel
                http://www.michaellarabel.com/

                Comment


                • #9
                  I propose tests with all Gameworks disabled. I must also warn to secure that the distance on some effects like Anisotropic and LOD are not reduced by the Nvidia driver. Another proposal that users have made in the past is to test Polaris and Vega undervolted by 100mv or so.

                  Comment


                  • #10
                    Originally posted by artivision View Post
                    test Polaris and Vega undervolted by 100mv or so
                    Nvidia overclocks too, you know. And overclocks better, too.
                    Last edited by xpue; 05-27-2018, 01:54 AM.

                    Comment

                    Working...
                    X