Announcement

Collapse
No announcement yet.

Where AMD's Catalyst Driver Does The Best On Linux

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

  • #11
    A Blender cycles rendering benchmark could be another option too...

    Comment


    • #12
      Originally posted by andrei_me View Post
      A Blender cycles rendering benchmark could be another option too...
      Blender's cycle engine OpenCL support is incomplete. No point using it as benchmark until it would support all features supported by CUDA implementation.

      Comment


      • #13
        Originally posted by dungeon View Post

        Per other Windows reviews i can't say Fury is best bang for buck when it comes to OpenCL performance as opossed to what happens to OpenGL... doubles OpenCL performance is even more slower then Hawaii which means 390/X is probably at best for the buck, of course if not some workstation FirePro cards
        Perhaps, I can't speak about the 290/390 series, I do know first hand that my two Fury X's vastly out produces my two 7970's running FAH. I went from 275k average to over 700k. Power usage increased 75w or so according to my ups.
        http://folding.extremeoverclocking.c...php?s=&u=13877

        Comment


        • #14
          Originally posted by blacknova View Post

          Blender's cycle engine OpenCL support is incomplete. No point using it as benchmark until it would support all features supported by CUDA implementation.
          Why? You wouldn't want to compare it to the CUDA version (if the features used were different) but there's no reason you couldn't test the OpenCL backend against different drivers. It works well enough to do so and be meaningful.

          Comment


          • #15
            Originally posted by smitty3268 View Post
            Why? You wouldn't want to compare it to the CUDA version (if the features used were different) but there's no reason you couldn't test the OpenCL backend against different drivers. It works well enough to do so and be meaningful.
            Actually, I would be very interested in seeing an Nvidia's CUDA vs AMD's OpenCL test with Blender Cycles. This would very much affect a Blender user's purchase decision.

            Comment


            • #16
              Originally posted by xeekei View Post
              Actually, I would be very interested in seeing an Nvidia's CUDA vs AMD's OpenCL test with Blender Cycles. This would very much affect a Blender user's purchase decision.
              I would too, if it was a fair comparison. If the CUDA render was higher quality, then it wouldn't be fair to compare them.

              AFAIK, the OpenCL code in Blender currently disables a couple of options because it can't handle them yet. I'm not sure if those are disabled in the CUDA backend as well, or not - or if they can optionally be disabled.

              Comment

              Working...
              X