AMD Radeon Linux Gaming Performance At Parity Between KDE Plasma 6.0 X11 vs. Wayland

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • mSparks
    Senior Member
    • Oct 2007
    • 2091

    #91
    Originally posted by qarium View Post

    no absolutly no one ever expected Xwayland to outperform X11... the only expectation in performance was native-wayland should outperform x11.

    Xwayland is legacy only code. no sane person should ever be interested in legacy-only code

    "and to hard to develop for"

    no one ever expected a easy-to-develop-for solution because they did know that as soon as they want to increase security it will automatically become hard-to-develop-for.

    any person who thinks you get security for free and this "easy" is an insane person.,
    That was the pitch for why time and money should be invested in wayland to replace X11, hindsight is a wonderful thing, see also

    For those picking up on this as "news", please read the following list: PCSX2 still supports Wayland. It just prefers the XCB/XWayland platform by default. You can set the I_WANT_A_BROKE...


    "xwayland is only legacy code"

    alternatively

    the reason X11 has the large and diverse userbase it has is because it has incorporated a large quantity of feedback from a large number of people with a diverse set of requirements over a very long period of time. Ditch any one of those requirements and you also ditch the users that want/need it.

    Comment

    • piotrj3
      Senior Member
      • Mar 2019
      • 839

      #92
      Originally posted by oiaohm View Post

      Guess I just got lucky; the run with --logverbose 6 did eventually crash the OS. Here’s the bug report log: http://kegel.com/linux/piglit-nvidia-bug-report.log.gz The last two lines from piglit were running: spec/arb_texture_rectangle/texrect_simple_arb_texrect [3524/4261] skip: 230, pass: 796, dmesg-warn: 1, fail: 2494, dmesg-fail: 2, crash: 1 tail -f /var/log/kern.log showed: Sep 16 09:58:14 rbb-ubu1604-3 kernel: [ 1475.867942] ext_texture_for[12354]: segfault at 0 ip 00007fcd50ebf640 sp...


      Turns out when you go and double check Nvidia claimed opengl support things go south quickly. Lot of ways Nvidia like Volkswagen before emissions scandal. They use work around to make themselves past and when you test without the work around they are not up to standard.

      Conformance doesn't mean supports everything perfectly, generally conformance means you support X, Y, Z etc. features, any significantly big piece of software will have errors. Here on phoronix we had AMD GPUs failing to play GTAV on Wine/Proton (nvidia was fine), RustiCL despite having excellent start, had to iron some bugs, etc. If you look at google you can find significant number of crashes during various workloads for Opengl, directx, Vulkan etc... There is too many examples. None of companies are perfect here.

      I am saying that something as simple as opengl calls implicit sync being not synced, would end up horribly in heavy majority of games using opengl. In Wayland present case it is simply Nvidia not signalling properly they didn't finish rendering yet and something reading not probably finished rendering buffer.

      Comment

      • piotrj3
        Senior Member
        • Mar 2019
        • 839

        #93
        Originally posted by qarium View Post

        what do you expect of Xwayland ? it is a legacy codepath... wait until Wine/Proton has wayland support then you see all the windows games in wayland mode and only then you see faster performance also keep in mind that the change from implicit sync to explicit sync also ingrease the performance a little.
        Yes, but there are 4 advantages Wayland possibly could have:

        1st. Xwayland doesn't support all things X does, so technically it might have certain performance advantages from simply having less features.
        2nd. X technically always puts all graphics calls into single queue. Modern GPUs do support execution of many queues. This could improve utilization. Even if Xwayland puts stuff in one queue all stuff outside of X (like DE, etc.) could be another so potentially DE is less blocking stuff from Xwayland.
        3rd. Modern GPUs have hardware accelerated scheduler. In windows known as HAGS. Now HAGS don't always improve performance, sometimes it makes it worse, but sometimes it improves quite a lot, potential area to investigate and improve performance again with.
        4th. Maybe utilization of SAM (smart access memory) or known as resizable BAR ( i know it is more driver level, but modern things should be made more aware of it).

        I know Xwayland's part of X won't have big performance increases - that is extremly unlikely. But Wayland's compositor probably should be more efficient at utilizing GPU comparing to X server composing. That is where that 1-2% i could expect.

        Comment

        • WileEPyote
          Senior Member
          • Nov 2023
          • 223

          #94
          Originally posted by mSparks View Post

          That was the pitch for why time and money should be invested in wayland to replace X11, hindsight is a wonderful thing, see also

          For those picking up on this as "news", please read the following list: PCSX2 still supports Wayland. It just prefers the XCB/XWayland platform by default. You can set the I_WANT_A_BROKE...


          "xwayland is only legacy code"

          alternatively

          the reason X11 has the large and diverse userbase it has is because it has incorporated a large quantity of feedback from a large number of people with a diverse set of requirements over a very long period of time. Ditch any one of those requirements and you also ditch the users that want/need it.
          Nothing is preventing someone else from taking over on x11 development. Nobody seems to want to.

          Comment

          • ezst036
            Senior Member
            • Feb 2018
            • 681

            #95
            Originally posted by mSparks View Post
            xservers repository is about as active as any other mature piece of software that works and is in widespread use.
            ​Clearly you didn't read what the discussion was. (Hint, it was a conversation about a person, an individual.)

            Originally posted by mSparks View Post
            xservers repository is about as active as any other mature piece of software that works and is in widespread use.



            Also, the whole point of OSS is also that its not dependant on any one entity or contributor, if something is broken - fixing it yourself and sending upstream the changes isn't just possible it is actively encouraged.

            Declaring an OSS project depreciated is about as meaningful as declaring the sun depreciated because you can only work the night shift - interesting talking point, but meaningless to anyone else.
            Read post 34 in its entirety, not just the convenient part. I already addressed this.

            Comment

            • ezst036
              Senior Member
              • Feb 2018
              • 681

              #96
              Originally posted by Weasel View Post
              I meant that he's the one making releases now, instead of Adam Jackson.


              Xorg (the X server running on hardware) has no maintainer, no-one committed to make releases, no-one willing to invest in it.​
              ???

              Comment

              • oiaohm
                Senior Member
                • Mar 2017
                • 8454

                #97
                Originally posted by piotrj3 View Post
                I am saying that something as simple as opengl calls implicit sync being not synced, would end up horribly in heavy majority of games using opengl. In Wayland present case it is simply Nvidia not signalling properly they didn't finish rendering yet and something reading not probably finished rendering buffer.
                Nvidia has a lot of detect games apply correction quirks that AMD and Intel don't need because of the very problem..

                Comment

                • Library
                  Junior Member
                  • Mar 2024
                  • 1

                  #98
                  The input lag in Wayland is the reason I refuse to use it when playing certain games. Turning off the compositor in X11 seems to work better for my use case.
                  ​I hope this changes.

                  It is specifically noticable in Rocket League played via Steam + Proton to me.

                  Comment

                  • Weasel
                    Senior Member
                    • Feb 2017
                    • 4503

                    #99
                    Originally posted by t1r0nama View Post
                    WTH? They are the most efficient way to use displays. In fact they are so efficient that even mac os and windows try to have same features as WMs on linux.

                    There is just small learning curve about 1-2 weeks and you'll get used to it and it will be harder for you to revert to DEs then it was switching to WMs.
                    No, it's going ass backwards to MS DOS era, closer to one app per screen bullshit. The less control the user has, the worse it is.

                    Tiling WMs are probably half the reason Wayland is in such garbage state, it seems it's all it cares about. Disgusting beyond belief.

                    Comment

                    • oiaohm
                      Senior Member
                      • Mar 2017
                      • 8454

                      Originally posted by Weasel View Post
                      No, it's going ass backwards to MS DOS era, closer to one app per screen bullshit. The less control the user has, the worse it is.

                      Tiling WMs are probably half the reason Wayland is in such garbage state, it seems it's all it cares about. Disgusting beyond belief.
                      One of the reason for zoned that noted in

                      Hello everyone! This is a new attempt to resolve the issues clients designed for stacking window managers are facing when they want to set their own...


                      Context for compositors: They now know the explicit layout of windows a client has created, and that these windows belong together, so can decide to e.g. allow the user to move them as a cluster between virtual zones, or represent them as one in a tiling WM and potentially only expand them if selected

                      This happens not to be limited to tiled.

                      Weasel how do you make sure a multi application end up correctly grouped on you taskbar on windows and X11? The answer is you don't. There is context information missing to make this happen correctly all the time.

                      One of the problems worked out is lack of context with multi process applications to know if they are multi processes or a application that should be stacked with each other in the taskbar.

                      As I said you need to read 264 there are lot issues that the existing X11/Windows solutions have had that have caused Wayland development in this area to stall out. Nothing todo with security.

                      Yes the represent them as one in a tiling WM is basically appear as a virtual desktop like window in the tiling WM tile zone is an option. So allowing multi applications in a single tile of a tiling windows manager how is this not better Weasel.

                      Weasel basically spend a little more time reading the line I quoted careful and then read 264 some more. The issue here is more complex than you think.

                      Something tiling WM makers are really want is context information so they are really good at finding where you are lacking context information so causing users minor to major issues. So most of the time when tiling WM developer says this has lack of context turns out this is messing up something else like task manager, taskbar and so on in your non tiling desktops.

                      Comment

                      Working...
                      X