Announcement

Collapse
No announcement yet.

XWayland 2D Performance Appears Better Than XMir

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

  • XWayland 2D Performance Appears Better Than XMir

    Phoronix: XWayland 2D Performance Appears Better Than XMir

    In the past few days I delivered X.Org vs. XMir Ubuntu Unity benchmarks on Intel hardware and Nouveau / NVIDIA. The benchmarking also found that 2D was also slower with XMir than simply running an X.Org Server. Benchmarks now carried out of X.Org vs. XWayland show that the Wayland-based equivalent is generally faster, at least for 2D operations...

    http://www.phoronix.com/vr.php?view=MTM5OTY

  • #2
    fanboys will be fanboys

    One could also speculate that wayland is significantly worse from the fact that he couldn't run anything that requires 3D and that for other tests weston broke with segfaults. Making such conclusions about which display server is superior on the basis of couple of tests is stupid and irresponsible.

    Comment


    • #3
      XMir has 1-2 missing features that caused performance problems and this has been known since Michael published his results. Doesn't mean Wayland is going to be better.

      Comment


      • #4
        Everyone predicted the results!

        Well, this is not a surprise!

        The benchmarks are done in Weston, which is Wayland native. Only the apps use XWayland.

        Wayland Programmers predicted these results: X Apps on a Wayland environment will be the about the same or faster than on a full X environment. (because it's like they are fullscreen: the X server does less calculus)


        The XMir benchmarks were done with a full X emulation via XMir: Unity 7 ran under XMir (not Mir native), so apps running under it are not fullscreen: can only be lower performance.

        Comment


        • #5
          Originally posted by mannerov View Post
          Well, this is not a surprise!

          The benchmarks are done in Weston, which is Wayland native. Only the apps use XWayland.

          Wayland Programmers predicted these results: X Apps on a Wayland environment will be the about the same or faster than on a full X environment. (because it's like they are fullscreen: the X server does less calculus)


          The XMir benchmarks were done with a full X emulation via XMir: Unity 7 ran under XMir (not Mir native), so apps running under it are not fullscreen: can only be lower performance.
          And you know why?
          Because XWayland was developed having in mind to run X legacy apps inside a Wayland world. It was not developed to runs an entire desktop environment on top of it.
          But Canonical needs testing for their Mir, so they will force every ubuntu users to become a tester.
          That is the difference: people that know what to do (Wayland devs) and a bunch of beginners (Mir devs).
          You are right, taking that into consideration, these (and future) results are predictable.

          Comment


          • #6
            Originally posted by d2kx View Post
            XMir has 1-2 missing features that caused performance problems and this has been known since Michael published his results. Doesn't mean Wayland is going to be better.
            No, Wayland is going to be better because it's developed by people who know what they're doing, instead of simply copying stuff from people who know what they're doing...

            Comment


            • #7
              Mir's superiority and awesomeness is secured at the protocol level:
              ...with our protocol- and platform-agnostic approach, we can make sure that we reach our goal of a consistent and beautiful user experience across platforms and device form factors. Canonical(C)
              So no matter what any benchmarks show - Mir is better than Wayland, period.

              Comment


              • #8
                Originally posted by Anarchy View Post
                One could also speculate that wayland is significantly worse from the fact that he couldn't run anything that requires 3D and that for other tests weston broke with segfaults. Making such conclusions about which display server is superior on the basis of couple of tests is stupid and irresponsible.
                One could, sure, if one were a ubuntu fanboy in denial...

                Can't wait for bo$$ to troll this thread and call the entire world basement dwellers.

                Comment


                • #9
                  Keep in mind that as far as I can tell xmir and xwayland are somewhat different.

                  xmir seems to me like its a fullscreen x server, that runs a full x session, including a full X window manager like compiz, that is forwarded to display on a mir server, and doesn't seem integrated.

                  xwayland on the other hand is ROOTLESS, in which x programs run, and Weston has its own miniature window manager that integrates the X apps with the wayland apps.

                  Comment


                  • #10
                    Originally posted by mannerov View Post
                    Well, this is not a surprise!

                    The benchmarks are done in Weston, which is Wayland native. Only the apps use XWayland.

                    Wayland Programmers predicted these results: X Apps on a Wayland environment will be the about the same or faster than on a full X environment. (because it's like they are fullscreen: the X server does less calculus)


                    The XMir benchmarks were done with a full X emulation via XMir: Unity 7 ran under XMir (not Mir native), so apps running under it are not fullscreen: can only be lower performance.
                    Good point. What I don't understand is why Ubuntu isn't going straight to Unity on Mir? I was under the impression that it was running fine, so why bother with xmir? (No troll replies please)

                    Comment


                    • #11
                      Originally posted by chrisb View Post
                      Good point. What I don't understand is why Ubuntu isn't going straight to Unity on Mir? I was under the impression that it was running fine, so why bother with xmir? (No troll replies please)
                      It seems Unity 8 is ready for Mir, but not Unity 7.

                      I don't know why they want to make it run under XMir. One can only make suppositions (didn't expect lower performance? Want to clear bugs of XMir by forcing users to test it? Marketing move? Want to give a message to AMD and Nvidia to hurry doing Mir driver?)

                      Comment


                      • #12
                        Originally posted by mark45 View Post
                        Mir's superiority and awesomeness is secured at the protocol level:

                        So no matter what any benchmarks show - Mir is better than Wayland, period.
                        L O L. I love how Canonical's bullshit marketing works. They post stupid youtube videos or some bs on some bs blogs and bunch of stupid kids are like "OW WOW AWESOME, CANONICAL ROX SOX".

                        Comment


                        • #13
                          Amusing reply to the "XFCE/LXDE/Gnome running on Xmir" videos from a Mir dev: wait a minute, thats a bug! Jokes aside, they are discussing some things in that thread that they want to deal with to improve performance (not that I would really understand the technical stuff discussed there, so read it yourself).

                          Comment


                          • #14
                            Originally posted by phoen1x View Post
                            and bunch of stupid kids are like "OW WOW AWESOME, CANONICAL ROX SOX".
                            And how is that any worse than bashing Canonical based on preliminary results from a few benchmarks of development snapshots?
                            I'm going to agree with Anarchy: drawing such bold conclusions regarding the superiority of one or another at this point is silly.
                            I wish both technologies the best and I'll make up my mind once I've run both on my computer.

                            Comment


                            • #15
                              Sadly the benchmarks were performed on a Ironlake (i3) CPU/GPU with IPS. As you can see the results with IPS are very unstable, the GPU is up/down clocked over a period of about 2 minutes and can vary in performance by about a factor of 2, along with side-effects from the CPU/memory clocks being independently controlled by the CPU C-states (and cpufreq). Along with the silliness of benchmarking with UXA, I'd take these benchmarks with a large grain of salt and would be very careful not to draw any conclusions about X vs Xwayland [vs Xmir].

                              Comment

                              Working...
                              X