Announcement

Collapse
No announcement yet.

17-Way NVIDIA Binary vs. AMD Open-Source Linux 4.6 / Mesa Git Driver Tests

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

  • 17-Way NVIDIA Binary vs. AMD Open-Source Linux 4.6 / Mesa Git Driver Tests

    Phoronix: 17-Way NVIDIA Binary vs. AMD Open-Source Linux 4.6 / Mesa Git Driver Tests

    Here is the continuation of yesterday's article that was a 10-way NVIDIA GPU Linux comparison with now having more NVIDIA results in plus also testing various AMD GCN GPUs using Linux 4.6.0 and Mesa Git...

    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
    Wow, I'm really impressed at how well AMD's open driver is competing against the nVidia blob

    Comment


    • #3
      We'll have to get you set up to run SHOC on the ROC stack. That should give some interesting open source numbers.
      Test signature

      Comment


      • #4
        Originally posted by bridgman View Post
        We'll have to get you set up to run SHOC on the ROC stack. That should give some interesting open source numbers.
        Has any good guide been made yet for fully setting up the ROC stack? Last I tried based upon GPUOpen.com info, etc, I ran into some library issues but didn't feel like messing around all day with it so just abandoned it for the time being.
        Michael Larabel
        https://www.michaellarabel.com/

        Comment


        • #5
          Originally posted by bridgman View Post
          We'll have to get you set up to run SHOC on the ROC stack. That should give some interesting open source numbers.
          Does that still only work with linux-4.4.0-kfd-compute-rocm-rel-1.1-15 ? Is all that open source?

          Comment


          • #6
            Originally posted by Michael View Post
            Has any good guide been made yet for fully setting up the ROC stack? Last I tried based upon GPUOpen.com info, etc, I ran into some library issues but didn't feel like messing around all day with it so just abandoned it for the time being.
            I thought the answer was yes, but sounds like we might need to take another look. Will ask...

            Originally posted by ernstp View Post
            Does that still only work with linux-4.4.0-kfd-compute-rocm-rel-1.1-15 ?
            So far yes, although we'll probably be updating to a 4.5-based kernel as the next step towards upstreaming (the 4.5-based internal tree has latest upstream radeon/amdgpu bits).

            Originally posted by ernstp View Post
            Is all that open source?
            I think so, but I'm not sure whether HCC is running over the HSAIL finalizer (not yet open) or the Lightning compiler (open) at this moment. Everything else should be open though, unless I forgot something.
            Last edited by bridgman; 31 May 2016, 07:10 PM.
            Test signature

            Comment


            • #7
              Originally posted by bridgman View Post
              I think so, but I'm not sure whether HCC is running over the HSAIL finalizer (not yet open) or the Lightning compiler (open) at this moment. Everything else should be open though, unless I forgot something.
              I'm running both Kaveri and Carrizo with HCC configured with LLVM git and it runs great.

              Comment


              • #8
                Performance per watt is interesting, but could you add Performance per USD? I'm certain you've done that before, but it's still quite interesting to see how much value we get per krona (or USD as you crazy Americans like to call it )

                Comment


                • #9
                  Originally posted by Azpegath View Post
                  Performance per watt is interesting, but could you add Performance per USD? I'm certain you've done that before, but it's still quite interesting to see how much value we get per krona (or USD as you crazy Americans like to call it )
                  PTS has support for determining perf-per-dollar, but it's unfortunately difficult to do that for cards well after launch.... Since there are many different cards out there for same model, do I use the old reference price? the average current price? the lowest price? Thus due to a lot of variables what to do for something besides launch pricing, don't tend to do it then.
                  Michael Larabel
                  https://www.michaellarabel.com/

                  Comment


                  • #10
                    Originally posted by Michael View Post

                    PTS has support for determining perf-per-dollar, but it's unfortunately difficult to do that for cards well after launch.... Since there are many different cards out there for same model, do I use the old reference price? the average current price? the lowest price? Thus due to a lot of variables what to do for something besides launch pricing, don't tend to do it then.
                    I understand that it would be a bit non-deterministic for older cards... But for me I think it would be OK just to have a ballpark figure about launch price. Like "which cards is this card supposed to compete against when it came out?". Mainly, I'm thinking about Nvidia vs AMD generations.

                    Comment

                    Working...
                    X