Announcement

Collapse
No announcement yet.

FreeBSD & DragonFlyBSD Put Up A Strong Fight On AMD's Threadripper 2990WX, Benchmarks Against Linux

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

  • FreeBSD & DragonFlyBSD Put Up A Strong Fight On AMD's Threadripper 2990WX, Benchmarks Against Linux

    Phoronix: FreeBSD & DragonFlyBSD Put Up A Strong Fight On AMD's Threadripper 2990WX, Benchmarks Against Linux

    The past two weeks I have been delivering a great deal of AMD Threadripper 2990WX benchmarks on Linux as well as some against Windows and Windows Server. But recently I got around to trying out some of the BSD operating systems on this 32-core / 64-thread processor to see how they would run and to see whether they would have similar scaling issues or not like we've seen on the Windows side against Linux. In this article are FreeBSD and DragonFlyBSD benchmarks with the X399 + 2990WX compared to a few Linux distributions.

    http://www.phoronix.com/vr.php?view=26770

  • #2
    Fascinating. With those first few benchmarks I thought FreeBSD and DragonFlyBSD were going to walk away with a massive victory. But then things more or less evened out. Impressive - my impression is that the *BSD contributor community is far smaller than the Linux contributor community, so their achievements are notable.

    Comment


    • #3
      Hey Michael, are you aware of a new EULA from Intel that is prohibitive on: "3. ... or (v) publish or provide any Materials benchmark or comparison test results."

      This is a license for a new firmware from Intel to fix those new vulnerabilities that cause impact on performance. The folks at Debian already rejected theses files to be officially packaged on their distro.

      People over Slashdot are getting ape shit mad about the situation: https://linux.slashdot.org/story/18/...filing-allowed
      Last edited by [email protected]; 08-23-2018, 08:49 AM.

      Comment


      • #4
        Originally posted by [email protected] View Post
        Hey Michael, are you aware of a new EULA from Intel that is prohibitive on: "3. ... or (v) publish or provide any Materials benchmark or comparison test results."

        This is a license for a new firmware from Intel to fix those new vulnerabilities that cause impact on performance. The folks at Debian already rejected theses files to be officially packaged on their distro.

        People over Slashdot are getting ape shit mad about the situation: https://linux.slashdot.org/story/18/...filing-allowed
        Yes I've known about it since last week but just seems to be getting attention today due to Reg / Slashdot... But it's not meaning no benchmarking on Intel CPUs period, it's about comparing the microcode revisions.... Anyhow, I am told not to worry too much and thus haven't bothered writing about it.
        Michael Larabel
        http://www.michaellarabel.com/

        Comment


        • #5
          Originally posted by Michael View Post

          Yes I've known about it since last week but just seems to be getting attention today due to Reg / Slashdot... But it's not meaning no benchmarking on Intel CPUs period, it's about comparing the microcode revisions.... Anyhow, I am told not to worry too much and thus haven't bothered writing about it.
          Be honest and imagine now that all software have such prohibition to not compare revisions (nothing before and after) what would you benchmark then?
          Last edited by dungeon; 08-23-2018, 11:10 AM.

          Comment


          • #6
            Originally posted by dungeon View Post

            Be honest and imagine now that all software have such prohibition to not compare revisions (nothing before and after) what would you benchmark then?
            Well first Intel have to prove that a non signed EULA is valid in the country that Michael resides in. None of the EULAs (even the written ones) are basically valid where I live for example.

            Comment


            • #7
              Originally posted by dungeon View Post

              Be honest and imagine now that all software have such prohibition to not compare revisions (nothing before and after) what would you benchmark then?
              That clause is going to be removed soon.... Just being blown out like wild on all these different sites today.
              Michael Larabel
              http://www.michaellarabel.com/

              Comment


              • #8
                What were the GCC versions on the Linux distros ? Also it was a shame that you couldn't compare against some of the more performant Linux distros like Clear.

                Comment


                • #9
                  Originally posted by Grinch View Post
                  What were the GCC versions on the Linux distros ? Also it was a shame that you couldn't compare against some of the more performant Linux distros like Clear.
                  GCC versions are shown on the system table (as always) within the article.

                  As said in the article, Clear isn't currently booting on that system. OpenSUSE Tumbleweed is quite performant on TR2 as shown in last round of tests.
                  Michael Larabel
                  http://www.michaellarabel.com/

                  Comment


                  • #10
                    Repeat benchmarks when BSD will have CPU vulnerabilities fixed or disable them for Linux.

                    FreeBSD:

                    Spectre v1, v2, v4 are vulnerable. Meltdown is disabled

                    https://wiki.freebsd.org/Speculative...ulnerabilities
                    Last edited by Pawlerson; 08-23-2018, 05:46 PM.

                    Comment

                    Working...
                    X