Announcement

Collapse
No announcement yet.

X.Org vs. XMir On KDE, Xfce, Unity Desktops

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

  • X.Org vs. XMir On KDE, Xfce, Unity Desktops

    Phoronix: X.Org vs. XMir On KDE, Xfce, Unity Desktops

    The latest interesting Linux test results to share this week for those not at Oktoberfest are 2D and 3D/OpenGL benchmark results when testing XMir and a pure X.Org Server configuration with the Xfce, Unity, and KDE desktops as will be found in next month's Ubuntu 13.10 release.

    http://www.phoronix.com/vr.php?view=19143

  • #2
    What's more interesting that XMirs performance hit
    is the huge performance hit by KDE.
    It's up to five times slower than Unity which runs Compiz!

    How the heck can it be that slow?

    Comment


    • #3
      XWayland

      I would have loved to see XWayland thrown into the mix.

      Comment


      • #4
        I think you should also benchmark app version that communicate with Mir/Wayland directly
        As there also exist Qt for wayland and QT for Mir , gtk for wayland

        Comment


        • #5
          Originally posted by uid313 View Post
          I would have loved to see XWayland thrown into the mix.
          Not possible because XWayland works on a rootless X environment while current XMir is the opposite

          Originally posted by miskol View Post
          I think you should also benchmark app version that communicate with Mir/Wayland directly
          As there also exist Qt for wayland and QT for Mir , gtk for wayland
          while for wayland you have weston and some other desktop environments been port to it, for Mir Unity 8 isnt ready yet so theres no easy possibility of testing that.

          Comment


          • #6
            Originally posted by uid313 View Post
            I would have loved to see XWayland thrown into the mix.
            For intel, Xmir uses sna, whereas XWayland uses uxa. Results are therefore predictable for these X drawings benchmarks.

            Wait for a glamor backend for XWayland (that would work on all cards), and the results will be more interesting.

            Comment


            • #7
              Originally posted by Pajn View Post
              What's more interesting that XMirs performance hit
              is the huge performance hit by KDE.
              It's up to five times slower than Unity which runs Compiz!

              How the heck can it be that slow?
              Previously people have blamed Oxygen style for KDE performance issues:

              Oxygen rendering path is much more complex than qt curve, due to gradients, animations, etc.
              It is therefore expected to be slower and this does not make it a bug. It is your call to decide between performances and features
              Also note that nvidia (or rather, at least some of the versions of the linux drivers) are known to behave badly with respect to cached pixmaps, which oxygen does alot (as opposed to QtCurve), in order to "improve" performances on other graphics card (and other drivers). Not much we can do about this.

              Comment


              • #8
                Now there is essentially no reason not to use XMir, other than driver support for EGL from AMD and NVIDIA. The benefits from using XMir instead of X are good enough reason to ditch X for now.

                Comment


                • #9
                  Originally posted by mmstick View Post
                  Now there is essentially no reason not to use XMir, other than driver support for EGL from AMD and NVIDIA. The benefits from using XMir instead of X are good enough reason to ditch X for now.
                  you do realise you need the x-stack to allow x-mir and x-wayland to do their thing ... ? :P

                  you'll be looking at black screens if you ditch X!

                  Comment


                  • #10
                    Meanwhile "LinuxGamer":

                    Last edited by verde; 09-24-2013, 12:11 PM.

                    Comment


                    • #11
                      Originally posted by mmstick View Post
                      Now there is essentially no reason not to use XMir, other than driver support for EGL from AMD and NVIDIA. The benefits from using XMir instead of X are good enough reason to ditch X for now.
                      Can you tell me about "the benefits"? Even though the performance is not a problem anymore (on Intel, we should check if that's valid with radeon and nouveau too), does it come with any reason to use it instead pure X.org? It is still an extra layer of bugs (not intended as a criticism, every software contains bugs), so I expect a real benefit from using it.

                      Comment


                      • #12
                        Originally posted by verde View Post
                        Meanwhile "LinuxGamer":
                        ROTFL!! But trolls always seem to find something to nag about. It's a strange gift they have.

                        This news is just wonderful. I was doubtful 13.10 was going to be useful for gaming and thought we would have to wait until 14.04, but this takes my doubts away.

                        XMir performance hit gone, SteamOS, and NVidia helping Nouveau developers. What a day.

                        Comment


                        • #13
                          Originally posted by mmstick View Post
                          Now there is essentially no reason not to use XMir, other than driver support for EGL from AMD and NVIDIA. The benefits from using XMir instead of X are good enough reason to ditch X for now.
                          What benefits are those, again? Still, good to see that they've at least got performance back up to par.

                          Comment


                          • #14
                            Good on Canonical, but these tests need to be given with AMD and NVIDIA graphics cards as well :P

                            I feel like wayyyy too many people are confusing XMir with Mir at this point. That really needs to be fixed...
                            XMir == X
                            They are pretty much one in the same (XMir is X running on top of Mir). With "composite bypass" for fullscreen OpenGL applications, of course the performance would be similar.

                            What I want to see is the performance for HD movies and such (windowed and fullscreen), as that is what I do with my computer, not games

                            I don't get the hate on Ubuntu just because of the Mir thing. I loved Unity until 12.04 when they completely destroyed it's performance on my computer...

                            Comment


                            • #15
                              Originally posted by Daktyl198 View Post
                              Good on Canonical, but these tests need to be given with AMD and NVIDIA graphics cards as well :P

                              I feel like wayyyy too many people are confusing XMir with Mir at this point. That really needs to be fixed...
                              XMir == X
                              They are pretty much one in the same (XMir is X running on top of Mir). With "composite bypass" for fullscreen OpenGL applications, of course the performance would be similar.

                              What I want to see is the performance for HD movies and such (windowed and fullscreen), as that is what I do with my computer, not games

                              I don't get the hate on Ubuntu just because of the Mir thing. I loved Unity until 12.04 when they completely destroyed it's performance on my computer...
                              For fullscreened movies wouldn't we expect the window manager to use overlays/composite bypass?

                              Comment

                              Working...
                              X