Announcement

Collapse
No announcement yet.

AMDGPU-PRO vs. NVIDIA OpenCL Performance With ArrayFire Using 18 GPUs

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

  • AMDGPU-PRO vs. NVIDIA OpenCL Performance With ArrayFire Using 18 GPUs

    Phoronix: AMDGPU-PRO vs. NVIDIA OpenCL Performance With ArrayFire Using 18 GPUs

    With now being able to benchmark ArrayFire via the Phoronix Test Suite, I've been having fun running a number of OpenCL graphics card tests with the 300+ available AF tests. The tests over the past week have been using the NVIDIA Linux driver while here are our first Radeon benchmark results using the AMDGPU-PRO 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
    Michael, do you know why the images/graphs are not loading in this article? I could see the first picture on the first page, the rest are missing on my side. Another thing, congrats on the 40k post mark!

    I tried reading the article. It feels like I'm watching TV with a blank screen and only the sound is coming through.

    Comment


    • #3
      Jabberwocky looks like openbenchmarking.org was down for a while

      Comment


      • #4
        Wow Fury trumped NVidia on FFTs 1d and 2d! I'm disappointed on AMD sort performance - it's worth looking into whether amd bolt is as slow as arrayfire's. I suspect the linear algebra routines needs tuning and carry Nvidia tuning-bias. Not sure what's up with Gfor, that's a weird routine anyway. I'll also note the ELWISE benchmarks seem mostly to go to NVidia - I suspect it's mainly due to newer compute architectures as it was mostly 980 and 1080s that beat out all the AMD options and the operations for ELWISE are supposed to be dirt simple. I'm actually quite surprised with those as I suspected they'd be memory-bandwidth dominated.

        Fantastic results and while the meaning may be dubious encour for ROCm and CUDA (not opencl)!.

        Comment


        • #5
          Why do we give a crap about AMD OpenCL on Linux until the stack is on par with Windows?

          Comment


          • #6
            Originally posted by nevion View Post
            Jabberwocky looks like openbenchmarking.org was down for a while
            Weird, I did not notice any 404s in my console. It's working now :-)

            Originally posted by Marc Driftmeyer View Post
            Why do we give a crap about AMD OpenCL on Linux until the stack is on par with Windows?
            The OpenGL and Vulkan stack is not on par with Windows either, yet we find it interesting to benchmark it and see how it improves over time or to see where there issues are. Yes, not everyone are using AMD OpenCL on Linux right now. This article is for the inquisitive people who are thinking of using it or for the few that already are.

            Comment


            • #7
              Is the AMDGPU-PRO driver as mature as the nVidia one for OpenCL?
              Hi

              Comment


              • #8
                It is, however, some OpenCL programs are quite sensitive to the platform they are programmed with. There are luxmark benchmarks in which AMD cards outperform the fastest NVidia ones. Furthermore, NVidia doesen't support OpenCL 2, which can give a nice performance boost.

                Comment


                • #9
                  No surprises here, memory bus width is king.

                  Comment


                  • #10
                    The results are all over the place. And not just for AMD, but in some tests the 780ti beats the 1080, while in others the 780ti is in last place while the 1080 is in first place. Not sure if these results are meaningful in any way...

                    Comment

                    Working...
                    X