Announcement

Collapse
No announcement yet.

Kicking Off April With An Eight-Way BSD/Linux Comparison

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

  • Kicking Off April With An Eight-Way BSD/Linux Comparison

    Phoronix: Kicking Off April With An Eight-Way BSD/Linux Comparison

    For getting April started, here is a fresh comparison of various BSDs and Linux distributions tested on an Intel Core i7 6800K Broadwell-E box. Tested operating systems included Antergos, Clear Linux, DragonFlyBSD 4.8, FreeBSD 11.0, Scientific Linux 7.3, TrueOS 20160322, Ubuntu 16.04.2 LTS, and Ubuntu 17.04 20170330.

    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
    For not being particularly optimized, Antergos scored quite well IMHO. I was always surprised why it scored so badly in previous tests.

    Comment


    • #3
      You can have bad luck with pstate driver of certain kernel versions.
      With acpi-cpufreq performance (and hopefully also schedutil), it likely would show a constant performance over a period of time.

      Comment


      • #4
        Finally we see final results

        Comment


        • #5
          Originally posted by phoronix View Post
          Phoronix: Kicking Off April With An Eight-Way BSD/Linux Comparison

          For getting April started, here is a fresh comparison of various BSDs and Linux distributions tested on an Intel Core i7 6800K Broadwell-E box. Tested operating systems included Antergos, Clear Linux, DragonFlyBSD 4.8, FreeBSD 11.0, Scientific Linux 7.3, TrueOS 20160322, Ubuntu 16.04.2 LTS, and Ubuntu 17.04 20170330.

          http://www.phoronix.com/vr.php?view=24477

          phoronix, some of the test graphs have compiler options listed in the bar itself, and some do not.

          I looked into the OpenBenchmarking.org results file, and saw that for ClearLinux the CFLAGS and CXXFLAGS environment variables include -O3, while sometimes the flags listed in the graph bars include -O3, and sometimes not.

          Whereas for other distributions, which do not specify C(XX)FLAGS env variables, the bars sometimes include -O2, and sometimes not, yet are more often empty.

          Could you please explain the three possibilities (no flags, flags specifying -O, flags not specifying -O) ?
          Reference link would be fine, of course.

          Comment


          • #6
            This is a GCC vs clang comparison, more than BSD vs linux. Still, it's amazing to see Dragonfly giving FreeBSD a run for its money!

            Comment


            • #7
              Ok, so thanks for the interesting benchmarks, but one thing baffles me, why the different optimization settings for the compiler on different OS'es ?

              This means what you are actually comparing is the performance of compilers at different settings, not the performance of the operating systems.

              As it stands, only the Go benchmarks are of any real worth in terms of relative operating system performance, the rest of the tests are rather worthless in this regard.

              Please, since these are very interesting benchmarks, could you redo them in a not so flawed way ?

              Using the same compiler options and preferably (although that's a bit more work) the same compilers on all system, -O2 on GCC is not the same as -O2 on Clang as far as which optimizations are enabled, so if you only test one option, please use -O3 .

              Also why were there no Go benchmarks for the BSD's ?

              Comment


              • #8
                Originally posted by jacob View Post
                Dragonfly giving FreeBSD a run for its money!
                "giving [...] a run for its money" just doesn't feel like the proper expression to use when talking about free software...

                Comment


                • #9
                  Michael, you should include Solus in a future benchmark to give us an idea on how that fares in common tasks and gaming.

                  Comment


                  • #10
                    How was Debian broken, I wonder?

                    Comment

                    Working...
                    X