Announcement

Collapse
No announcement yet.

7-Way Linux Desktop Gaming Comparison On Ubuntu 14.10

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

  • 7-Way Linux Desktop Gaming Comparison On Ubuntu 14.10

    Phoronix: 7-Way Linux Desktop Gaming Comparison On Ubuntu 14.10

    With Ubuntu 14.10 bringing various updates to the packaged desktop environments along with updated open-source graphics drivers, here's our per-cycle usual tests of the popular Linux desktops while looking at their impact on the Linux gaming performance.

    http://www.phoronix.com/vr.php?view=20924

  • #2
    Nice to see KDE isn't at the bottom in every test anymore. Overall it seemed to be the most average, since its highs weren't that high and its lows weren't that low. Lets hope it continues to improve as they continue shifting toward KF5.

    Comment


    • #3
      I only miss mate here.
      It's available in straight-away from official repositories in Ubuntu 14.10, no ppas needed.

      Comment


      • #4
        On the second page, after the first OpenArena v0.8.8 graph, there's a mistake:

        With the OpenArena open-source game, the performance between the different desktop environments were minimal with the Intel Haswell graphics on the stock Ubuntu 14.10 setup.
        It should be:

        With the OpenArena open-source game, the performance differences between the different desktop environments were minimal with the Intel Haswell graphics on the stock Ubuntu 14.10 setup.
        Aside from that, it seems that only with a bare startx can the full performance be got. How are desktop environments in the way of games? Is it about the resources they use, or are there any other reasons?

        Comment


        • #5
          Wow, that Gnome Shell hit Counterstrike took looks extremely interesting! I have only been running my Steam games within Gnome Shell (3.10 or 3.12, both I think) and haven't given it a thought that the performance might be bad.

          I know it's a hassle for you to test Steam Games, but is it possible for you to test it out with some more games? Only compare Gnome Shell with, let's say, XFCE (since it was pretty stable), and run 4-5 games on it?

          I also agree with kalrish that you should try testing it with startx.

          Comment


          • #6
            Well these results make no sense? first E17 is the fastest, then it?s the slowest? Same for other tests/DEs. I would like to know what?s going on.

            (The good thing is Xfce, which I?m using, is never slow.)

            Comment


            • #7
              There must be something wrong with your setup Michael, on my machine with 7770 and mesa git I get pretty much same framerate under Unity as under Gnome 3.12. However I use Gnome Staging PPA which has updated packages for thing like mutter etc (Ubuntu ships with some old version), maybe it has something to do with it.

              Comment


              • #8
                Looks like no compositing penalty for modern graphics

                This jibes with my recent tests comparing IceWM with Cinnamon on my AMD/Radeon machines, where no game shows a performance difference even though Cinnamon has a reputation for being extremely heavy. I am guessing this is because games use so much more resources than any DE, no matter how heavy, and possibly the use of a similar code path for OpenGL compositing and for the OpenGL game itself.

                Responsiveness of opening a new window or that sort of thing is quite another matter, and in some versions of Cinnamon a bug specific to the menu can make the whole DE seem too slow for something like a 2 GHZ Pentium 4. On really old or resource limited machines, video playback takes a hit if any kind of compositing DE is used. Neither a Intel Atom N455 nor a Pentium 4 2GHZ can run Cinnamon or compiz and play 720p/30fps H264 video at the same time with no lag, both can play that same video just fine with a non-compositing window manager. This is of course CPU playback where no GPU decoding is available. For any machine with a PCI-e slot, a modern video card, no matter how small, makes 1080p playback over a compositing window manager trivial, and I set those machines up with my full featured OS, dialing back only on desktop effects that can slow opening new windows.

                Pentium 4's are great for seeing just how heavy a desktop environment REALLY is! I never realized in the old days (2008-2010) how much capacity compiz ate, because I didn't have any 720p video or other HD video to worry about and compiz of that era was much lighter than now. On modern desktops I'd say don't worry about the overweight nature of Unity, GNOME 3, or Cinnamon (a gnome-shell fork), as you will never see a difference in CPU heat, GPU heat, or power usage and responsiveness is fine. For laptops it varies: small ones and expecially netbooks pick up battery life when running a noncompositing desktop, possibly by letting the openGL parts of the GPU shut down, possibly by just reducing the load if 2d acceleration is over the 3d hardware.

                Comment


                • #9
                  It should be noted that while basically every DE hit minimums of about 30-35 fps, Unity stayed above 50 fps.

                  I also would have liked to see Openbox as well. Really though, a separate Wayland/XWayland Mir/XMir comparison where he takes Ubuntu running Unity 8 and whatever top 3 distros are running Wayland and see how they do be more popular amoung the readers here than throwing in Openbox. This gives a good idea in performance bars expected anyways.
                  Last edited by profoundWHALE; 09-19-2014, 02:23 PM.

                  Comment


                  • #10
                    Originally posted by profoundWHALE View Post
                    It should be noted that while basically every DE hit minimums of about 30-35 fps, Unity stayed above 50 fps.
                    I'm not sure what you're trying to say here. Unity clearly hit 30 fps on one test, and 15 fps on another. Are you just looking at a single test in particular?

                    Comment

                    Working...
                    X