Announcement

Collapse
No announcement yet.

NVIDIA 24-Way GPU Comparison With Many OpenCL, CUDA Workloads

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

  • #11
    Originally posted by eoyilmaz View Post
    Hey Michael, another great benchmark again.

    BTW, I highly suggest you to include Redshift3D to your benchmarks. It has a commandline benchmark tool. Oh, it is CUDA only for now, but they just recently started to support AMD cards in OSX with Metal API and I bet that they will extend it on all OSes soon (obviously not with Metal).
    Do you have any info on the benchmark commands / documentation? Some time ago I looked into the demo but I believe at least then it didn't have proper command line automation.
    Michael Larabel
    http://www.michaellarabel.com/

    Comment


    • #12
      Originally posted by onlyLinuxLuvUBack View Post


      Hello Michael,

      I feel you produce soo much information that I can't even keep up with it.
      Have you thought of taking the data after the article(you have already done all the work) and then doing a youtube video going over it.
      I was thinking maybe you would get money from advertisers on youtube or bring more traffic to this site?
      Thought about it, but YouTube video production can be very time consuming.
      Michael Larabel
      http://www.michaellarabel.com/

      Comment


      • #13
        Originally posted by zexelon View Post
        Sadly OpenSource is not into ML and CUDA type workloads. They keep trying but cant quite catch up. ROCm is the best so far perhaps... but its going to take a couple of years to reach where CUDA is now.
        Serious ML farms tend to run Linux clusters with an opensource software stack + CUDA/cuDNN.
        If you mean only graphics cards, AMD's pro drivers are not open-source as well..

        Comment


        • #14
          Originally posted by zexelon View Post
          Sadly OpenSource is not into ML and CUDA type workloads. They keep trying but cant quite catch up. ROCm is the best so far perhaps... but its going to take a couple of years to reach where CUDA is now.
          Serious ML farms tend to run Linux clusters with an opensource software stack + CUDA/cuDNN.
          If you mean only graphics cards, AMD's pro drivers are not open-source as well..

          Comment


          • #15
            mppix yah thats what i meant. A lot of the AMD fans on this site tend to forget that the AMD pro drivers are still closed anyway.

            At my company, we used a linux OSS stack with nvidia hardware and CUDA.

            Comment


            • #16
              Originally posted by eoyilmaz View Post
              Hey Michael, another great benchmark again.

              BTW, I highly suggest you to include Redshift3D to your benchmarks. It has a commandline benchmark tool. Oh, it is CUDA only for now, but they just recently started to support AMD cards in OSX with Metal API and I bet that they will extend it on all OSes soon (obviously not with Metal).
              Looked back at it... Not fond of its Linux handling due to some hard-coded path restrictions thus requiring root access to install due to relying on INI files in /usr/redshift even if overriding the install path.

              But anyways, added it. More details @ https://twitter.com/phoronix/status/1305262168535961602
              Michael Larabel
              http://www.michaellarabel.com/

              Comment


              • #17
                Originally posted by Michael View Post

                Looked back at it... Not fond of its Linux handling due to some hard-coded path restrictions thus requiring root access to install due to relying on INI files in /usr/redshift even if overriding the install path.

                But anyways, added it. More details @ https://twitter.com/phoronix/status/1305262168535961602
                This is great, thank you very much, appreciated your efforts.

                Andyes, Redshift runs faster in Linux. More interestingly it runs even faster when you have multiple GPUs, especially GeForce cards vs Quadro cards. Because Geforce cards can run in headless mode under Linux where as (afaik) they can not under Windows (Quadro cards can do that with Tesla Compute Cluster (TCC) driver) and there is the WDM overhead of Windows that you can not run away if the card is not in headless mode.

                Comment

                Working...
                X