Announcement

Collapse
No announcement yet.

Google Cloud Tau T2A Ampere Altra vs. T2D AMD EPYC Performance

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

  • Google Cloud Tau T2A Ampere Altra vs. T2D AMD EPYC Performance

    Phoronix: Google Cloud Tau T2A Ampere Altra vs. T2D AMD EPYC Performance

    Over the summer Google announced Tau T2A as the first Google Cloud Compute Engine VM to run on Arm. The T2A series is powered by Ampere Altra processors and complement the Tau T2D series powered by AMD EPYC 7003 "Milan" processors that launched last year. I have been trying out the Tau T2A series for the past several weeks and have some initial benchmarks to share today for showing how the Ampere Altra Arm VMs can perform against the existing T2D series.

    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
    Thanks for these benchmarks!
    I think though that the comparison would be more enlightening if you provide graphs that show the price on the x-axis.

    Comment


    • #3
      x86 destroys ARM once again, "software ecosystem issue".
      ARM is competitive with x86, or slightly ahead, "omg look how great ARM is!!!! x86 is legacy/dead!"

      I swear Michael is just like the Russian propagandists when it comes to hardware and software instead. Laughable bias against facts.

      Comment


      • #4
        EPYC Zen 4 is programmed for the coming months? EPYC Zen 3 came out over a year after Zen 3, perhaps 18 months even, will Zen 4 be that early?

        Comment


        • #5
          Originally posted by Weasel View Post
          I swear Michael is just like the Western propagandists when it comes to hardware and software instead. Laughable bias against facts.
          I have fixed it for you.

          Comment


          • #6
            Originally posted by Setif View Post
            I have fixed it for you.
            Here's a gold example: https://t.co/cGqebwwjYx

            Translation:
            [Numbers are claims of Russian official]
            losses of the Ukrainian army - about 100,000 people
            losses of the Russian army - 5,937 people
            therefore Russia needs to mobilize 300,000 people
            Sounds legit.

            I swear if this was a movie it would probably be categorized into "comedy". Unfortunately it's real, with real consequences on (russian) people.

            Comment


            • #7
              Originally posted by Weasel View Post
              Here's a gold example: https://t.co/cGqebwwjYx

              Translation:Sounds legit.

              I swear if this was a movie it would probably be categorized into "comedy". Unfortunately it's real, with real consequences on (russian) people.
              Apparently protestors against russia's mobilization will be forced to serve - another one from the Nazi playbook. Giving your opponents guns can only end well...

              Comment


              • #8
                It's always good to see more than one ISA that's competing. That said, I wonder how Xeon compares to those two.

                In many of the tests I noticed a much larger curve between Ampere Altra 8 cores and 16 cores, compared to AMD Epic 8 cores and 16 cores. Could that be single thread performance / IPC impact?

                Comment


                • #9
                  Originally posted by Jabberwocky View Post
                  It's always good to see more than one ISA that's competing. That said, I wonder how Xeon compares to those two.

                  In many of the tests I noticed a much larger curve between Ampere Altra 8 cores and 16 cores, compared to AMD Epic 8 cores and 16 cores. Could that be single thread performance / IPC impact?
                  EPYC likely can't run at max turbo with 16+ active cores. Also the halving of 32MB L3 per core is a large effect from 8 to 16 cores. Altra doesn't have similar turbo behaviour and has very little L3 to start with, so scales better with core count.

                  Comment

                  Working...
                  X