Announcement

Collapse
No announcement yet.

The Performance Impact Of Spectre Mitigation On POWER9

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

  • The Performance Impact Of Spectre Mitigation On POWER9

    Phoronix: The Performance Impact Of Spectre Mitigation On POWER9

    Over the past year we have looked extensively at the performance impact of Spectre mitigations on x86_64 CPUs but now with having the Raptor Talos II in our labs, here are some benchmarks to see the performance impact of IBM's varying levels of Spectre mitigation for POWER9.

    http://www.phoronix.com/vr.php?view=27070

  • #2
    Now that's fucking huge. However I'd be interested in knowing QEMU performance for x86 applications on power9, especially 3D intensive one.

    Comment


    • #3
      Yikes... and I thought Intel got hit hard...

      Comment


      • #4
        Michael, many thanks for reviewing this very interesting machine, i'm seriously considering getting either Lite or Blackbird for home by the end of the year.

        It would be interesting to know how the different SMT modes affect Spectre mitigation on Power9, if you could find the time to re-do some benchmarks with SMT off ( ppc64_cpu --smt=off ). A nice thing of these machines is that you can setup the SMT behavior on the go, per core. This, and process binning can be very useful to optimize some workloads with specific single threaded processes

        Comment


        • #5
          It would be interesting to see some gaming benchmarks. Not because I think it'll be any good at it, but because I want to see how far off it is (with both x86 emulation and native code).

          Comment


          • #6
            Originally posted by Djhg2000 View Post
            It would be interesting to see some gaming benchmarks. Not because I think it'll be any good at it, but because I want to see how far off it is (with both x86 emulation and native code).
            That would be interesting, especially natives FOSS games so you can actually test them in both native and qemu x86. Also please compare to an x86 cpu.
            ## VGA ##
            AMD: X1950XTX, HD3870, HD5870
            Intel: GMA45, HD3000 (Core i5 2500K)

            Comment


            • #7
              Originally posted by Djhg2000 View Post
              It would be interesting to see some gaming benchmarks. Not because I think it'll be any good at it, but because I want to see how far off it is (with both x86 emulation and native code).
              I don't think this is a thing, as far as i know Wine/Proton can not work on PPC or ARM and current native games are x86 only and as far as i know Power cannot emulate x86 with hardware acceleration(the same way X86 host qemu cannot use KVM for AARCH64 for example), so the only possible scenario i think is possible is an open source game like xonotic that could compile on PPC unpatched

              Comment


              • #8
                Originally posted by jrch2k8 View Post

                I don't think this is a thing, as far as i know Wine/Proton can not work on PPC or ARM and current native games are x86 only and as far as i know Power cannot emulate x86 with hardware acceleration(the same way X86 host qemu cannot use KVM for AARCH64 for example), so the only possible scenario i think is possible is an open source game like xonotic that could compile on PPC unpatched
                QEMU can translate binaries as you run them. ExaGear is leveraging this (with their own performance modifications) for running x86 games on ARM.

                Comment


                • #9
                  Originally posted by Djhg2000 View Post

                  QEMU can translate binaries as you run them. ExaGear is leveraging this (with their own performance modifications) for running x86 games on ARM.
                  Yeah, i guess my English got in the way, what i meant is right now there is no hardware acceleration for cross architecture emulation/virtualization hence the IPC penalty will be massive enough to make any game test worthless.

                  Also i'm not sure if it is even possible to emulate x86_64 on PPC with hardware acceleration since IBM don't have an x86 license deal, hence it can open the doors for Intel to sue the living bejesus out of them(i read this several times on forums searching around for x86 emulation on PPC but i'm not 100% this is actually legit but sounds about right as your usual technology being held back due to legal crap)

                  Comment


                  • #10
                    Originally posted by jrch2k8 View Post

                    Yeah, i guess my English got in the way, what i meant is right now there is no hardware acceleration for cross architecture emulation/virtualization hence the IPC penalty will be massive enough to make any game test worthless.

                    Also i'm not sure if it is even possible to emulate x86_64 on PPC with hardware acceleration since IBM don't have an x86 license deal, hence it can open the doors for Intel to sue the living bejesus out of them(i read this several times on forums searching around for x86 emulation on PPC but i'm not 100% this is actually legit but sounds about right as your usual technology being held back due to legal crap)
                    Ah, I see. Indeed, it will be a major bottleneck for modern games. Really old games like CS 1.6 or Carmageddon might be playable though, so I figured it could be worth a try.

                    Worst case we get to see the potential an x86 to POWER9 recompiler could have (when comparing native vs x86 versions of games like OpenArena).

                    Comment

                    Working...
                    X