Announcement

Collapse
No announcement yet.

AMD EPYC 7401P: 24 Cores / 48 Threads At Just Over $1000

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

  • #11
    I'd love to see 7601X2 :-)

    Comment


    • #12
      It would have been nice to do the HPC benchmarks on these server CPUs.

      Comment


      • #13
        Originally posted by eddy View Post
        Hmm, so Epyc does badly here because the PostgreSQL test is not NUMA aware?
        Maybe, or maybe it's just because it uses NUMA. Databases benefit a lot of from fast access to RAM and I/O, which is a problem with these chips.

        Comment


        • #14
          Originally posted by nils_ View Post

          Maybe, or maybe it's just because it uses NUMA. Databases benefit a lot of from fast access to RAM and I/O, which is a problem with these chips.
          Yes, "NUMA aware" isn't a very good term. The benchmark seems to be just firing hundreds of SQL queries at a database, which might saturate the infinity fabric. Depends on how big the database is though and how it's setup.

          Comment


          • #15
            Michael, I would be very interested to see how there EPYC parts compare to the Threadripper platform. I am soon going to be building myself a beefy personal server machine (not for corporate/enterprise needs) to run CPU intensive tasks on and for large storage with ZFS. I am trying to decide between Threadripper and EPYC, particularly since Threadripper also supports ECC RAM, unlike Intel's consumer CPUs.

            The highest-end Threadripper is priced similarly to the 24-core EPYC you reviewed here ($999 vs $1075 IIRC), so it would be very interesting to see how Threadripper's higher clocks, but fewer cores, cache and memory channels compare to EPYC's more cores with more memory channels and cache, but lower clocks.

            This article would have been perfect if it included the Threadripper 1950x. The raw performance and performance/cost graphs are both very interesting. Is there any way to meaningfully combine results from your different benchmarks (possibly using some OpenBenchmarking/PTS tool?) to get the info I want? Or would that be meaningless due to variations in the test setups / the other hardware?

            BTW, just bought a new phoronixpremium subscription. Please make it happen.

            Comment


            • #16
              Originally posted by tajjada View Post
              Michael, I would be very interested to see how there EPYC parts compare to the Threadripper platform. I am soon going to be building myself a beefy personal server machine (not for corporate/enterprise needs) to run CPU intensive tasks on and for large storage with ZFS. I am trying to decide between Threadripper and EPYC, particularly since Threadripper also supports ECC RAM, unlike Intel's consumer CPUs.

              The highest-end Threadripper is priced similarly to the 24-core EPYC you reviewed here ($999 vs $1075 IIRC), so it would be very interesting to see how Threadripper's higher clocks, but fewer cores, cache and memory channels compare to EPYC's more cores with more memory channels and cache, but lower clocks.

              This article would have been perfect if it included the Threadripper 1950x. The raw performance and performance/cost graphs are both very interesting. Is there any way to meaningfully combine results from your different benchmarks (possibly using some OpenBenchmarking/PTS tool?) to get the info I want? Or would that be meaningless due to variations in the test setups / the other hardware?

              BTW, just bought a new phoronixpremium subscription. Please make it happen.
              AMD requested I not run any side-by-side tests of Threadripper and EPYC due to intended for different markets :/ But those knowing how to use OpenBenchmarking.org can likely generate a similar comparison with a few clicks.
              Michael Larabel
              https://www.michaellarabel.com/

              Comment


              • #17
                Originally posted by Michael View Post

                AMD requested I not run any side-by-side tests of Threadripper and EPYC due to intended for different markets :/ But those knowing how to use OpenBenchmarking.org can likely generate a similar comparison with a few clicks.
                Ah, fair enough. Didn't know. Now I understand why you haven't done such comparisons already.

                Yes, in that case, it is better to follow AMD's requests. They were generous enough to provide you with all the CPU samples, after all.

                Thanks for explaining the reason. I'll try to figure out how to make use of OpenBenchmarking.org to get the info I want.

                Comment


                • #18
                  Originally posted by Michael View Post
                  AMD requested I not run any side-by-side tests of Threadripper and EPYC due to intended for different markets
                  But you could compare Epyc against Core i7 7900X and 7980X, or did AMD request that you not do that either?

                  Originally posted by Michael View Post
                  intended for different markets
                  I think the markets are segmented by price more than anything else. AMD would be wise to allow comparisons of Epyc 7401P against Threadripper 1950X and Core i7 7900X, and 7551P against 7980X especially as they have no competing Threadripper product for the latter. I'm sure the Epycs will do well in sufficiently multithreaded tests.

                  Originally posted by tajjada View Post
                  I'll try to figure out how to make use of OpenBenchmarking.org to get the info I want.
                  Try this link:
                  OpenBenchmarking.org, Phoronix Test Suite, Linux benchmarking, automated benchmarking, benchmarking results, benchmarking repository, open source benchmarking, benchmarking test profiles

                  Comment


                  • #19
                    Originally posted by chithanh View Post
                    But you could compare Epyc against Core i7 7900X and 7980X, or did AMD request that you not do that either?
                    I will have 79xx comparison shortly.... Didn't have time yet.
                    Michael Larabel
                    https://www.michaellarabel.com/

                    Comment


                    • #20
                      Originally posted by chithanh View Post
                      Thank you. Exactly what I wanted.

                      Looks like the difference between TR 1950x and EPYC 7401P is quite small in most tests. I think I will go for the TR. This is a personal system, so I don't care about enterprise support/certification/etc. Motherboard will likely be cheaper, too. TR is also overclockable.

                      With a small overclock on the TR, I think I will be able to have similar or better multi-core performance on the TR's 16 cores as with EPYC's 24, while also getting much higher single-threaded performance from the high clocks. Again, this is a personal system, not enterprise, so I don't mind overclocking.

                      Comment

                      Working...
                      X