Announcement

Collapse
No announcement yet.

RadeonSI Primitive Culling Yields Mixed Benchmark Results

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

  • RadeonSI Primitive Culling Yields Mixed Benchmark Results

    Phoronix: RadeonSI Primitive Culling Yields Mixed Benchmark Results

    Yesterday's patches introducing RadeonSI primitive culling via async compute yielded promising initial results, at least for the ParaView workstation application. I've been running some tests of this new functionality since yesterday and have some initial results to share on Polaris and Vega...

    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
    Aww

    (you should have added NVIDIA in the ParaView tests to see whether we beat TITAN RTX or not...)

    Comment


    • #3
      @Marek: Your current system specs (apart from Vega20 (VII) monster ;-))?

      Comment


      • #4
        By it's design it stands to reason GpuTest would produce lower results. Some of these tests woulds naturally become less by what culling does.

        Comment


        • #5
          Well, that was expected. I mean, does ParaView cull at all?
          It was clear that games that do reasonable culling wouldn't benefit that much (or at all), otherwise both AMD and Nvidia would've implemented implicit compute culling into their drivers years ago.

          Of course, it's still a nice improvement for somebody who has to work with bad software.

          Regarding the differences to Mareks results, likely he didn't measure in UHD.
          Last edited by juno; 14 February 2019, 04:10 PM.

          Comment


          • #6
            So, glxgears run at 16K fullscreen is slower

            Comment


            • #7
              Originally posted by nuetzel View Post
              @Marek: Your current system specs (apart from Vega20 (VII) monster ;-))?
              I don't have a desktop computer at home. I have Ryzen 7 1700X in the office.

              Comment


              • #8
                Thx for the test Michael!
                If Marek could tweak his patches a little more we might see some small gains but this doesn't look like this is something to enable by default.
                Maybe there is a game out there which does profit from this but has yet to be found.

                Comment


                • #9
                  Or games have a bottleneck elsewhere.

                  Comment


                  • #10
                    When I played Serious Sam 3 it was CPU bottlenecked.
                    In Dota 2 there is very little to nothing what to cull, it's by design and it's also CPU bottlenecked.
                    Unigine - i'm pretty sure it does culling by itself.

                    Comment

                    Working...
                    X