Announcement

Collapse
No announcement yet.

6-Way Linux Desktop 2D/3D Performance Comparison On Ubuntu 14.04 LTS

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

  • 6-Way Linux Desktop 2D/3D Performance Comparison On Ubuntu 14.04 LTS

    Phoronix: 6-Way Linux Desktop 2D/3D Performance Comparison On Ubuntu 14.04 LTS

    Our latest benchmarking of Ubuntu 14.04 is looking at the 2D X11 and 3D OpenGL performance of six different desktop environments when benchmarked on Ubuntu 14.04 LTS. The tested desktops included Unity, Xfce, KDE, LXDE, GNOME Shell, and Openbox.

    http://www.phoronix.com/vr.php?view=19998

  • #2
    Interesting results.

    Since KDE seems to be the odd one out, is there any chance you could test openSUSE 13.1 KDE to see how it fares?


    Thanks.

    Comment


    • #3
      Originally posted by imperfectlink View Post
      Since KDE seems to be the odd one out, is there any chance you could test openSUSE 13.1 KDE to see how it fares?
      At some point in the future depending upon how many others request such a test...
      Michael Larabel
      http://www.michaellarabel.com/

      Comment


      • #4
        KDE Performance

        Desktop effects in KDE were disabled?

        Comment


        • #5
          What is the default configuration for KDE: Xrender or OpenGL? Which OpenGL version? Native/Raster? And what is the Vsync option?

          Since it impacts the performance, it would be interesting to have it with the tests.

          Comment


          • #6
            It always intrigued me how some tests perform better in composited, worse in non-composited and vise versa. It makes sense, but it really shows how there isn't a perfect graphics arrangement, depending on what you do. But, if you've got multiple GPUs, one configured for desktop use and the other configured for GPGPU tasks, I guess you get the best of both worlds.

            It's a shame KDE is still behind, I wonder if the wayland ports of it will get any better, or if maybe there's 1 small tweak that makes a big difference (such as vsync). I'm well aware KDE performs the worst, but my hardware is good enough that it doesn't really have enough of an impact on me for me to complain about.

            Comment


            • #7
              Originally posted by schmidtbag View Post
              It always intrigued me how some tests perform better in composited, worse in non-composited and vise versa. It makes sense, but it really shows how there isn't a perfect graphics arrangement, depending on what you do. But, if you've got multiple GPUs, one configured for desktop use and the other configured for GPGPU tasks, I guess you get the best of both worlds.

              It's a shame KDE is still behind, I wonder if the wayland ports of it will get any better, or if maybe there's 1 small tweak that makes a big difference (such as vsync). I'm well aware KDE performs the worst, but my hardware is good enough that it doesn't really have enough of an impact on me for me to complain about.
              I would assume 'Suspend Desktop effects for fullscreen windows" was not checked, as usual. If this is the case, I would suspect that enabling that would bring back in line with the other DEs.

              Comment


              • #8
                The things is. This test was testing the defaults. And I suspect the default is still the stupid non-redirect for fullscreen applications. (I don't know for sure since I haven't used kde in a year, xfce) Is there anyone who could confirm if this is the case or not?

                Comment


                • #9
                  Originally posted by ua=42 View Post
                  The things is. This test was testing the defaults.
                  Defaults on Ubuntu, which is infamous for intentionally breaking KDE.

                  Comment


                  • #10
                    Originally posted by schmidtbag View Post
                    It always intrigued me how some tests perform better in composited, worse in non-composited and vise versa. It makes sense, but it really shows how there isn't a perfect graphics arrangement, depending on what you do. But, if you've got multiple GPUs, one configured for desktop use and the other configured for GPGPU tasks, I guess you get the best of both worlds.

                    It's a shame KDE is still behind, I wonder if the wayland ports of it will get any better, or if maybe there's 1 small tweak that makes a big difference (such as vsync). I'm well aware KDE performs the worst, but my hardware is good enough that it doesn't really have enough of an impact on me for me to complain about.
                    KDE doesn't un-redirect by default for fullscreen apps, which imho, is the right option. If you enable that tickbox, games typically run a lot faster. Apparently the redirecting penalty on wayland can be pretty much zero :-)

                    Comment


                    • #11
                      So once again Micheal has not benchmarked actual performance, but performance under defaults of Ubuntu developers. Once again a useless benchmark, since the DEs/WMs were not run with same settings. A sad performance for a person running a benchmark site.

                      Comment


                      • #12
                        Originally posted by Vim_User View Post
                        So once again Micheal has not benchmarked actual performance, but performance under defaults of Ubuntu developers. Once again a useless benchmark, since the DEs/WMs were not run with same settings. A sad performance for a person running a benchmark site.
                        It isn't a useless benchmark - benchmarks are the MOST relevant when they're as close to the average setup as you can get. These benchmarks obviously don't show the true potential of each DE, but it takes hours to properly tweak and configure just 1 DE to perform optimally. No matter how many optimizations Michael makes, there's always going to be some nerd who says "dude you forgot to X your Y, this test doesn't tell us anything" when it might only give a 0.1% difference. The purpose of these benchmarks was to see which DE performs the best. If you optimize all of them equally, generally speaking, they should proportionally all increase at approximately the same rate, proving absolutely nothing different. So, that makes such optimizations a waste of time in the end.

                        There is no such thing as satisfying every user unless you spend an entire day benchmarking a single DE showing varying setups. KDE and GNOME were the only environments that should have had another separate set of results - one with compositing and another without.
                        Last edited by schmidtbag; 03-12-2014, 11:33 AM.

                        Comment


                        • #13
                          Originally posted by schmidtbag View Post
                          It isn't a useless benchmark - benchmarks are the MOST relevant when they're as close to the average setup as you can get. These benchmarks obviously don't show the true potential of each DE, but it takes hours to properly tweak and configure just 1 DE to perform optimally. No matter how many optimizations Michael makes, there's always going to be some nerd who says "dude you forgot to X your Y, this test doesn't tell us anything" when it might only give a 0.1% difference. The purpose of these benchmarks was to see which DE performs the best. If you optimize all of them equally, generally speaking, they should proportionally all increase at approximately the same rate, proving absolutely nothing different. So, that makes such optimizations a waste of time in the end.

                          There is no such thing as satisfying every user unless you spend an entire day benchmarking a single DE showing varying setups. KDE and GNOME were the only environments that should have had another separate set of results - one with compositing and another without.
                          I don't see it as on optimization to enable one feature in a DE to bring it up to par with the other DEs. Running KDE with un-redirect disabled compared with all others having that enabled tells absolutely nothing about actual performance, it only tells us that Ubuntu developers can't set sane defaults for KDE and that Michael once again is not able or willing to provide fair benchmarks.

                          Comment


                          • #14
                            Originally posted by bakgwailo View Post
                            I would assume 'Suspend Desktop effects for fullscreen windows" was not checked, as usual. If this is the case, I would suspect that enabling that would bring back in line with the other DEs.
                            Last I saw, that checkbox had "if (intel) return;" due to some old bugs. Not sure if it's fixed yet or not.

                            Comment


                            • #15
                              These tests are missing memory usage...

                              As for unredirect, i often play games in a maximized window (not fullscreen), so that one (vertical) panel shows me net status, speed, temp, and other info.

                              Comment

                              Working...
                              X