Announcement

Collapse
No announcement yet.

AMD Threadripper 2990WX Linux Benchmarks: The 32-Core / 64-Thread Beast

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

  • AMD Threadripper 2990WX Linux Benchmarks: The 32-Core / 64-Thread Beast

    Phoronix: AMD Threadripper 2990WX Linux Benchmarks: The 32-Core / 64-Thread Beast

    Whether you are compiling a lot of code, rendering models with Blender, or running various scientific workloads with OpenMP or MPI, the AMD Threadripper 2990WX is capable of delivering immersive Linux performance with its 32-cores and 64 total threads. While coming in at $1800 USD, the AMD Threadripper 2990WX can deliver better performance than the more expensive Intel Core i9 7980XE. Beyond being mesmerized about the performance today with this high-end desktop/workstation processor with the many thread-happy Linux workloads we encounter daily, this 32-core Zen+ processor has us even more eager to see AMD's next-generation Zen2-based EPYC CPUs next year.

    http://www.phoronix.com/vr.php?view=26700

  • #2
    Awesome review Michael, so much benchy goodness! One additional metric I would have liked to see, is what the idle power consumption was of the i9 vs old TR vs new TR. For those of us who leave their workstations running 24/7, we don't care that much about power consumption under load, because it's going towards doing real work. But power consumption at idle is important because the machine spends a lot of time idling and that's power that is wasted. Especially if you have multiple rigs. Of course there will be variables like different mobo's, but just seeing the overall idle consumption and how the top players all compare would be really useful.

    Comment


    • #3
      Damn... That's one really high performance CPU.

      Where I work we've been so impressed with the original Threadripper-series that it's replaced the Xeons we previously offered to our customers as part of our full software+hardware package (together with a really expensive 4k stereo monitor). Unless Intel comes up with something really crazy, and I don't mean anything like their oh-sh*t-we-need-to-cobble-together-something-really-cool-ASAP demo powered by a 1000W water cooler at Computex this year, I think they've basically lost the business from us for the next couple of years.

      Comment


      • #4
        That performed a lot better than I was expecting. It was pretty consistently faster than the 7890XE in every test except single-threaded, and anybody who is shopping for CPUs like these is not bound to care about stuff like that. Even the PPW results were a bit surprising.

        Comment


        • #5
          Any chance we can get some ffmpeg H.264 and H.265 encode testing? How about 7z or tar+pxz compression testing? Video encoding and compression are the main reasons I have for having a wide CPU.

          Nice review overall.

          Comment


          • #6
            Originally posted by Particle View Post
            Any chance we can get some ffmpeg H.264 and H.265 encode testing? How about 7z or tar+pxz compression testing? Video encoding and compression are the main reasons I have for having a wide CPU.

            Nice review overall.
            There will be x264/ffmpeg tests in upcoming articles like the Linux vs BSD and I think also Linux distro comparison tests. 7-zip should be in there too. I don't have any pxz test profile but if you provide a bash script example of what you'd be interested in I could add it to PTS (or any other tests you'd like to see).
            Michael Larabel
            http://www.michaellarabel.com/

            Comment


            • #7
              Nice review Michael.

              I wish I had the cash for a CPU like that...

              I just wondered about the benchmark selection, I haven't gone into all the details, but over at Anandtech, they frown abit upon the 2990wx for being bottlenecked when used with multicore memory intensive programmes. Is any of the test performed in this review using a combination of being multicore and memory intensive?

              Kind regards
              Brut.

              Comment


              • #8
                Originally posted by Particle View Post
                Any chance we can get some ffmpeg H.264 and H.265 encode testing? How about 7z or tar+pxz compression testing? Video encoding and compression are the main reasons I have for having a wide CPU.

                Nice review overall.
                +1 for video encoding benchmarks; that's my main use case for my upcoming build. Still a great review regardless!

                Comment


                • #9
                  What about results in Linux gaming? I don't know of any native Linux games able to use so many cores, but maybe some DX11 titles on Wine + DXVK could benefit of them (GTA5 is on my mind specifically).

                  Comment


                  • #10
                    Originally posted by Particle View Post
                    Any chance we can get some ffmpeg H.264 and H.265 encode testing?
                    Second that. However, the problem with video encoding tests on multi-core behemoths like Threadrippers and server processors is that encoders (at least x264, x265, and libvpx) don't scale all very well past a certain number of cores/threads. So with these kinds of processors the video encoding benchmarks currently available in the Phoronix Test Suite won't actually be pushing the CPU to its limits and taking full advantage of it as you'll be running into the limitations of the encoding software.

                    A better method would be running multiple encodes at a time and measuring CPU & memory usage and some other things so one could look at these numbers and make an educated guess on what's actually bottlenecking the process. A not-so-great practice with video encoding benchmarks is not reporting what the CPU usage was during the benchmark, because realistically if the utilisation is less than 50% then in a real-life scenario you'd probably be running two or more encodes at a time to get the most out of your hardware.

                    Encoding efficiency (as in, compression achieved per watt) also goes down the more threads you use, at least with x264, so that's something to consider as well. I also seem to remember that using too many threads with x264 can actually reduce video quality noticeably but by default x264 limits the maximum amount of threads so that doesn't happen.

                    Comment

                    Working...
                    X