Announcement

Collapse
No announcement yet.

AMD EPYC 9684X Genoa-X Provides Incredible HPC Performance

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

  • #11
    Could you add a 2nd geomean graph which excludes the HBM-only Xeon and adds back the benchmarks which failed on it?

    Comment


    • #12
      Michael Thanks for a great review that demonstrates the areas where the new Genoa-X CPUs accelerate. We're entering a time where there are specialized tools (CPUs) for specific workloads.
      I am looking forward to additional reviews that show the performance on more everyday tasks, such as database benchmarks. Given the obvious impact of HBM and L3cache in this area a series of runs with different data / user volumes would be incredible insightful.
      Hope you can squeeze such tests in before you inevitably have to return all of that juicy hardware :-)

      Comment


      • #13
        Good Lord, the absolute stomping Intel get in the benchmarks I care about for work related things is frightening.

        If only our AMD server didn't have so many issues, I'd be asking how much a maxed out Genoa-X server would cost... but our Intel boxes have not given us a single problem, while the AMD Milan server has gone back three times for repeated issues.

        Comment


        • #14
          Really? Sounds like you need to change vendors. I could say the same, but with Intel/amd roles reversed.

          Comment


          • #15
            Originally posted by pegasus View Post
            Really? Sounds like you need to change vendors. I could say the same, but with Intel/amd roles reversed.
            Did you switch vendors?

            Honestly, choice here is pretty limited. Our supplier is great for everything else, but that particular system just seems to be cursed. Others with similar systems report no issues. Hopefully the last return solved it (cross fingers, it's only been a few weeks) but from the list of stuff they replaced if it isn't solved this time I'm going to be significantly less circumspect in expressing my displeasure.

            Comment


            • #16
              Originally posted by Paradigm Shifter View Post
              but that particular system just seems to be cursed.
              This happened to me before, it's like Murphy's law magnet of issues, everything that could be faulty in the whole month of manufactured servers were concentrated in that single server that had to be one owned by me 😂


              Comment


              • #17
                The article says that the open foam test was too big but in this article not only was the medium mesh run but the large mesh (which would be the more interesting comparison)

                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

                Comment


                • #18
                  I should clarify “too big for HBM only”

                  Comment

                  Working...
                  X