Announcement

Collapse
No announcement yet.

Pentium G4600 vs. Ryzen 3 1200 On Ubuntu 17.10 Linux Benchmarks

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

  • Pentium G4600 vs. Ryzen 3 1200 On Ubuntu 17.10 Linux Benchmarks

    Phoronix: Pentium G4600 vs. Ryzen 3 1200 On Ubuntu 17.10 Linux Benchmarks

    Earlier this week I posted some benchmarks looking at Intel Pentium vs. AMD Ryzen 3 performance for Linux gaming. Those tests on the Pentium and Ryzen systems were done with both NVIDIA and AMD Radeon graphics for seeing how the gaming performance compares in the spectrum of sub-$100 CPUs and cheap graphics cards. But for those that were just curious about the CPU performance, here are some benchmarks I also carried out with the Pentium G4600 Kabylake and AMD Ryzen 3 1200.

    http://www.phoronix.com/vr.php?view=25934

  • #2
    That's a pretty cool article. It makes a specific point that compliments the prior article well.

    Comment


    • #3
      Thank you @Michael!
      It underpins what most argued about the former article.

      BTW
      I'm running latest amd-staging-drm-next (#0a7d506e54ca) on RX580 (Polaris20):

      amdgpu-pci-0100
      Adapter: PCI adapter
      vddgfx: +0.75 V
      fan1: 904 RPM
      temp1: +25.0°C (crit = +94.0°C, hyst = -273.1°C)
      power1: 34.00 W

      Last edited by nuetzel; 01-25-2018, 08:56 PM.

      Comment


      • #4
        With very AVX heavy workloads, the Pentium G4600 does much better.
        Ha, ha, Himeno was much faster on any Intel CPUs even before that AVX2 switch

        I don't think that is because of avx2, on Intel there -O3 switch makes most of difference and on AMD nothing AFAIR

        edit: and i also think stastic build was fastest on AMD, not sure about Intel.... tried that years ago
        Last edited by dungeon; 01-25-2018, 11:29 PM.

        Comment


        • #5
          Originally posted by dungeon View Post

          Ha, ha, Himeno was much faster on any Intel CPUs even before that AVX2 switch

          I don't think that is because of avx2, on Intel there -O3 switch makes most of difference and on AMD nothing AFAIR
          So what do you have to say about this, then:

          x264 v2017-09-08, asmFish v2017-09-19, C-Ray v1.1., Darktable v2.2.5 1+2 (real apps)?

          Clock and IPC can't be it... --- Ho, ho, ho.

          Comment


          • #6
            Originally posted by nuetzel View Post

            So what do you have to say about this, then:

            x264 v2017-09-08, asmFish v2017-09-19, C-Ray v1.1., Darktable v2.2.5 1+2 (real apps)?

            Clock and IPC can't be it... --- Ho, ho, ho.
            Here there is no GPU driver involved. Here compiler switches could involve some difference

            Try for example Himeno i mentioned by changing flags and you will see On Intel CPU -O3 makes much difference from -O2, while on AMD CPU -O3 was actually a bit slower than -O2 AFAIR

            It is far from just raw clocks and ipc of these CPUs who makes difference

            edit: can't remember, on AMD CPU from dynamic -O2 -mtune=native was fastest or something like that 15% more or something... so some instruction came in there and makes diff, but if you invoke -O3 instead it became a bit slower actually

            But that was years ago, maybe something magically is changed now with compilers, so try it and you'll see. Also march=native maded things slower also, only mtune should be used to be faster... there was some deep shit comipler shit right there
            Last edited by dungeon; 01-26-2018, 01:04 AM.

            Comment


            • #7
              With very AVX heavy workloads, the Pentium G4600 does much better.
              That's an interesting statement since the Pentium doesn't support AVX.

              Comment


              • #8
                Originally posted by BlackArchon View Post
                That's an interesting statement since the Pentium doesn't support AVX.
                Ah

                Comment


                • #9
                  Considering the lower clock of the Ryzen CPU even ST performance doesn't look bad for AMD.

                  Comment


                  • #10
                    Originally posted by GruenSein View Post
                    Considering the lower clock of the Ryzen CPU even ST performance doesn't look bad for AMD.
                    Yeah, if anything, it highlights that it's really good all-rounder. The FP/vector unit lacks compared to the kaby lake core, shown by e.g. himeno, but AMD will hopefully address that in a future generation.

                    Comment

                    Working...
                    X