Announcement

Collapse
No announcement yet.

16-Way Linux OS Performance Comparison

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

  • 16-Way Linux OS Performance Comparison

    Phoronix: 16-Way Linux OS Performance Comparison

    Building on our earlier 11-Way Linux/BSD Platform Comparison, starting a new week we're up to a 16-Way Linux operating system comparison. Added in now are results from PCLinuxOS, ROSA, the lightweight antiX distribution, and then the Gentoo-based Sabayon and Calculate Linux Desktop distributions.

    http://www.phoronix.com/vr.php?view=18809

  • #2
    PCLinuxOS Gaming

    Does anyone know what PCLinuxOS does to make their gaming so much better?

    Comment


    • #3
      Originally posted by caryhartline View Post
      Does anyone know what PCLinuxOS does to make their gaming so much better?
      Proprietary Catalyst driver by default while other distributions doesn't have it by default. Nothing too fancy.

      So if you install Catalyst Driver on other Distributions, you'll get similar outcome.

      Comment


      • #4
        Some times it helps to read the article:
        PCLinuxOS 2013 activates the proprietary Catalyst driver by default, rather than the open-source Radeon Gallium3D stack, which is why that distribution is showing the best OpenGL gaming results.

        Comment


        • #5
          Nice performance comparison Michael . I really like that LXDE holds up continuously through this tests and I must say whatever Ubuntu does for Unity and Compiz it really shows.I might even try it these days.
          Last edited by kUrb1a; 06-24-2013, 12:49 PM.

          Comment


          • #6
            Originally posted by caryhartline View Post
            Does anyone know what PCLinuxOS does to make their gaming so much better?
            RTFA "PCLinuxOS 2013 activates the proprietary Catalyst driver by default, rather than the open-source Radeon Gallium3D stack, which is why that distribution is showing the best OpenGL gaming results."

            Comment


            • #7
              The Gentoo-based Calculate Linux Desktop and Sabayon Linux distributions didn't offer any compelling performance advantages over the non-Gentoo distributions, contrary to some Gentoo users believing it's a magical speed demon.
              That's not how it works. The speed boost on Gentoo comes from compiling the packages with optimised compiler flags. PTS explicitly compares programs compiled with the same compiler flags, so it's obvious there will be no performance improvement there. The only thing that could mean a performance difference is the libraries that the tests rely upon being compiled with optimal settings (and we don't know if they were).

              Comment


              • #8
                Did you recompile Sabayon with -march=native or did you use the binary packages?

                Just asking, because if you did the former, there is no reason at all why Sabayon should have any speed advantage.

                Comment


                • #9
                  | The Gentoo-based Calculate Linux Desktop and Sabayon Linux distributions didn't offer any compelling performance advantages over the non-Gentoo distributions, contrary to some Gentoo users believing it's a magical speed demon.

                  I didn't appreciate this comment. Calculate and Sabayon are both BINARY-based distros that are Gentoo-based which means they have no real specific optimizations for hardware or USE flags or anything. For example, the GCC CFLAGS are optimized to work on everything as old as an Athlon 64 system from a decade ago.

                  If you REALLY want to attempt to make this argument, please test with a proper Gentoo (or Funtoo) installation that *IS* optimized for your hardware.

                  Also, as a sidenote, people who use Gentoo for the performance gain are generally using it for the wrong reason. Gentoo should be used because it is extremely customizable in terms of package versions and dependencies and USE flags and such, not because of the performance gain (unless you're a bank or similar where you need microsecond precision for transactions and such). For most users, the performance gain of using Gentoo itself is quite negated by the compile times anyway.

                  Comment


                  • #10
                    Originally posted by jdhore View Post
                    Also, as a sidenote, people who use Gentoo for the performance gain are generally using it for the wrong reason. Gentoo should be used because it is extremely customizable in terms of package versions and dependencies and USE flags and such, not because of the performance gain (unless you're a bank or similar where you need microsecond precision for transactions and such). For most users, the performance gain of using Gentoo itself is quite negated by the compile times anyway.
                    Not if it's a low-end computer (think Atom or AMD Brazos), where such speed increases are visible, and not if you have a distcc/icecream server set up

                    Comment


                    • #11
                      Two things that come to my mind: this is not a 16 way Linux OS comparison, since it also features PC-BSD, DragonflyBSD and Debian Gnu/kFreeBSD, those are not Linux by the very definition.

                      Also, can we have the graphs like any other site on the web I know of does it? If there is no result for a given OS in a specific benchmark than don't quietly remove it from the graph, leave it in without a result or use a minus sign to show that there is no result. Also, explain why those OSes have no result for that specific benchmark. I can guess that many of the games simply don't run on the OSes with BSD kernel, but why are they omitted in PostgreSQL and other benchmarks? Why is Debian GNU/kFreeBSD missing in some of the benchmarks? Why are there only 12 results in the Tachyon benchmark, where are the other four?

                      Come on, don't just post the results, explain why there are results missing!

                      Comment


                      • #12
                        Originally posted by GreatEmerald View Post
                        Not if it's a low-end computer (think Atom or AMD Brazos), where such speed increases are visible, and not if you have a distcc/icecream server set up
                        Fair point.

                        Comment


                        • #13
                          Originally posted by AJenbo View Post
                          Some times it helps to read the article:
                          It's easy to miss something with the way the pages are chopped up.

                          Comment


                          • #14
                            Originally posted by Vim_User View Post
                            Come on, don't just post the results, explain why there are results missing!

                            Maybe it is self explainable because one can see he is runing LLVMPipe on both Debian variants.

                            http://openbenchmarking.org/embed.ph...ha=9d684ce&p=2


                            Maybe he doing it that way because firmware is not loaded by default for that radeon card in Debian linux and with Debian kfreebsd variant because hardware mesa dri is not available at all .

                            So these are "defaulted" benchmarks - what one would see if he/she does not touch anything, it is - just load & run .
                            Last edited by dungeon; 06-24-2013, 02:50 PM.

                            Comment


                            • #15
                              Originally posted by Vim_User View Post
                              Two things that come to my mind: this is not a 16 way Linux OS comparison
                              There is a similar mistake in the closing paragraph:
                              There you have it, our 16-way Linux/BSD performance comparison. While there were 14 Linux distributions (and two BSDs) tested
                              I count 13 GNU/Linux distributions, two BSD distributions, and GNU/kFreeBSD which I suppose is neither...

                              Originally posted by Vim_User View Post
                              Why is Debian GNU/kFreeBSD missing in some of the benchmarks?
                              It probably means the Phoronix Test Suite was unable to build/run that test on that system, yet. So whereas PostgreSQL is available as a Debian package, it might have patches applied, build configuration adjusted to make it work; if the PTS were to try to build its own copy of that program, or some other version, it probably won't even compile. This is particularly likely on new or less common distributions.

                              In other cases it may be due to some missing feature like 3D acceleration, either not properly set up or not available at all. That might result in a pants score or simply not being able to run at all.

                              Comment

                              Working...
                              X