Announcement

Collapse
No announcement yet.

Ubuntu 20.04 GNOME X.Org vs. Wayland Session Performance Impact For Gaming

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

  • Originally posted by jacob View Post
    You are contradicting yourself. You can't support something and replace it at the same time. ...
    No, you're just limiting your own comprehension. Of course can one support it and replace it. The support is what is necessary for it to replace it. So you keep splitting hairs for which nobody cares. You could equally argue that it is slow, because that what it is supposed to be. And while you may be right does only nobody care for it. And until you get this, what it is and what people want it for, will nobody care for an interpretation as to why it fails to deliver. So you can talk about it forever in circles, just nobody will listen anymore or care for it when all it does for the is to throttle the performance, even when it's just games.

    Comment


    • Originally posted by sdack View Post
      No, you're just limiting your own comprehension. Of course can one support it and replace it. The support is what is necessary for it to replace it. So you keep splitting hairs for which nobody cares. You could equally argue that it is slow, because that what it is supposed to be. And while you may be right does only nobody care for it. And until you get this, what it is and what people want it for, will nobody care for an interpretation as to why it fails to deliver. So you can talk about it forever in circles, just nobody will listen anymore or care for it when all it does for the is to throttle the performance, even when it's just games.
      You're talking out of you arse. Wayland was never meant to "support" X11 because it was never meant to be a next-version X11 or even a successor to X11. It was meant to be a radical, clean sheet design, totally (and perhaps deliberately) incompatible with X11. And so it is.

      And if you have some actual hard facts supporting the claim that Wayland is somehow inherently slow (never mind the question how an API specification by itself can be slow) then please share them; saying "it's sloooow because I say so" doesn't impress anyone.

      Comment


      • Originally posted by jacob View Post
        You're talking out of you arse. Wayland was never meant to "support" X11 because it was never meant to be a next-version X11 or even a successor to X11. It was meant to be a radical, clean sheet design, totally (and perhaps deliberately) incompatible with X11. And so it is.
        Like I said, you're making excuses. There is a reason why Phoronix is making these benchmarks and you're not getting it. Or more likely do you refuse to get it, because you cannot accept that people want to run their games under Wayland and want to see them run faster. So you stick your head into the sand and come with excuses. You're trying to buy time. How many more years will it take?

        That Wayland is different from X11 was never a secret, but the driving idea behind it. You only want to sell this now as something new, but it's always been like this. It's sad that you there even try to claim that it wasn't meant to succeed X11. It was always meant to do this. Perhaps you're trying to convince yourself of some imaginary scenario, where there is a good reason why it doesn't perform as well as it should, but no matter how different you imagine Wayland is is it still a windowing system and it needs to perform fast or it won't find support. It doesn't matter how different or equal it is to other existing windowing systems. It needs to perform better and not worse. So keep making excuses, explain yourself how much others like myself misinterpret the results, but it's just you who cannot accept it.
        Last edited by sdack; 03-31-2020, 07:03 PM.

        Comment


        • Originally posted by sdack View Post
          Like I said, you're making excuses. There is a reason why Phoronix is making these benchmarks and you're not getting it. Or more likely do you refuse to get it, because you cannot accept that people want to run their games under Wayland and want to see them run faster. So you stick your head into the sand and come with excuses. You're trying to buy time. How many more years will it take?
          Well, it seems that unlike you I actually read and understand those benchmarks. Firstly the benchmark we are currently discussing here is about legacy X11 apps running under Wayland, not native Wayland apps; secondly it shows that with one single exception, there is no discernable difference in performance between that and native X11.
          People always want to see their software run faster, in the present case it means that once that software stops relying on the X11 emulation layer it will run faster than it does now, which will also be faster than it runs today on pure X11.

          You were saying?

          Comment


          • Originally posted by sdack View Post
            Like I said, you're making excuses. There is a reason why Phoronix is making these benchmarks and you're not getting it. Or more likely do you refuse to get it, because you cannot accept that people want to run their games under Wayland and want to see them run faster. So you stick your head into the sand and come with excuses. You're trying to buy time. How many more years will it take?

            That Wayland is different from X11 was never a secret, but the driving idea behind it. You only want to sell this now as something new, but it's always been like this. It's sad that you there even try to claim that it wasn't meant to succeed X11. It was always meant to do this. Perhaps you're trying to convince yourself of some imaginary scenario, where there is a good reason why it doesn't perform as well as it should, but no matter how different you imagine Wayland would be is it still a windowing system and it needs to perform fast or it won't find support. It doesn't matter how different or equal it is to other existing windowing systems. It needs to perform better and not worse.

            Really its not that simple. The benchmarks Phoronix did were all games using xwayland there are games with SDL backend that you can switch the backend wayland so avoid the xwayland overhead so in fact go faster.

            Also most of those benchmarks are inside the run to run error rate as in not more than 1 frame difference in performance also some of the benchmarks xWayland benchmark was faster than X11 server on bare metal.

            Please note some of us have benched SDL games on wayland where you can change the backend to wayland. There is something interesting. There is a bug in X11 protocol that results in everything being at least 1 frame late. Please note I said at least it could be 2 to 3 frames late. This is something wayland protocol fixes.

            So lets say wayland does work out at 1FPS slower but the person seeing under wayland that frame on screen 1 frame sooner than the person using X11 be it X11 bare metal or xwayland. That one frame per second slower in fact means nothing because the person using the wayland backend is getting to see the frames on time.

            Other than the lack of variable refresh rate under wayland if a game has a direct Wayland backed you will play better if you choose to use wayland. It comes a debate when game only X11 backend if xwayland or X11 bare metal is better.

            sdack you need to remember these benchmarks are xwayland on top of wayland you are reading. So if that is inside striking distance where do you think us using SDL based games that have a wayland backend in performance are in fact. So yes X.org X11 server is slow. Xwayland is slightly slower to equal and games on wayland itself are the fastest.

            Wayland has delivered it promise just legacy support is not going to be faster. But xwayland is getting legacy support on Wayland inside native speed by optimising stuff out of xwayland it does not need.

            Reality like it or not sdack current wayland compositors are not performing worse than X11 servers.

            Comment


            • Originally posted by jacob View Post

              Well, it seems that unlike you I actually read and understand those benchmarks. Firstly the benchmark we are currently discussing here is about legacy X11 apps running under Wayland, not native Wayland apps; secondly it shows that with one single exception, there is no discernable difference in performance between that and native X11.
              People always want to see their software run faster, in the present case it means that once that software stops relying on the X11 emulation layer it will run faster than it does now, which will also be faster than it runs today on pure X11.

              You were saying?
              No, there is nothing "legacy" about running games on Linux. Wayland would have to succeed X11 for games to become "legacy X11 apps", and according to you is this not the case when you say Wayland isn't supposed to support X11 (when in fact it is). You're not going to claim Wayland has turned Windows apps into "legacy Windows apps", or do you?

              So it seems while you're still trying to make up your mind about what Wayland is, do you in fact want Wayland to be the successor to X11 when you give X11 apps the "legacy" status. Your problem only is that Wayland isn't a suitable successor and that we can only dream of a future where this is different, while we've been having this dream for more than 10+ years now and these benchmarks keep telling us that it remains a dream.

              So who wants to use Wayland when the most demanding applications still run better under old X11? Only the dreamers do.

              Comment


              • Originally posted by oiaohm View Post
                sdack you need to remember these benchmarks are xwayland on top of wayland you are reading...
                Oh, I know. And you're adding more excuses to the pile. Nothing is changing these benchmark results.

                Comment


                • Originally posted by sdack View Post
                  No, there is nothing "legacy" about running games on Linux. Wayland would have to succeed X11 for games to become "legacy X11 apps", and according to you is this not the case when you say Wayland isn't supposed to support X11 (when in fact it is). You're not going to claim Wayland has turned Windows apps into "legacy Windows apps", or do you?

                  So it seems while you're still trying to make up your mind about what Wayland is, do you in fact want Wayland to be the successor to X11 when you give X11 apps the "legacy" status. Your problem only is that Wayland isn't a suitable successor and that we can only dream of a future where this is different, while we've been having this dream for more than 10+ years now and these benchmarks keep telling us that it remains a dream.

                  So who wants to use Wayland when the most demanding applications still run better under old X11? Only the dreamers do.
                  This is pure drivel. Wayland has nothing to do with Windows and it has nothing to do with X11 either. It's a graphics stack, not something to run "on X11". There is exactly 0 lines of code in the entire Wayland ecosystem that are related in any way whatsoever to X11, except for XWayland which is to Wayland what eXceed is (used to be?) to Windows or XQuartz is to MacOS.

                  There is nothing legacy about games on Linux but running anything X11-based on Linux (games included) is legacy. You don't whine about MacOS's Quartz being bad because X11 apps don't run as well as native apps on it, do you? Or do you? In fact as this benchmark shows there is basically no downside to running legacy X11 stuff on a Wayland-enabled system through XWayland, which is actually a pretty fantastic achievement. MacOS certainly doesn't come anywhere close to that in this regard. And I still don't get what you are whining about, precisely because as this benchmark shows, even the most demanding legacy X11 software (except for one) runs on top of Wayland with virtually no loss in performance or functionality.

                  Comment


                  • so i've spent a good 24 hours using gnome on wayland now. overall i have to say, its been a pleasant experience. a few quirks, but not sure if those are gnome or wayland specific. for example, in wayland mode, if i enable the maximize button via gnome tweaks, it has delay when i click it with some applications. its really odd. like gnome terminal, if i click to maximize it takes about 10 seconds to do it. if i click to un-maximize it takes another 10 seconds to go back. but here is the odd thing, if i right click the title bar and click maximize its instant. same with going back to un-maximize. if i double click to maximize / un-maximize its instant as well. but i don't have this oddity in xorg gnome.

                    another bug that i can solely blame on either wayland itself or just straight up compositor is epic games store severe lag. to the point its unusable completely unless you like waiting 20 - 70 seconds for EVERY action the store does. i mean everything. from loading up photos of your games, to slowly sliding up notifications, to clicking, launching a game, to trying to move the window itself or even clicking its client decorations. like the exit button, etc. i can't blame gnome for this because i tested it in sway too. same exact behavior. its similar behavior to the maximize button oddity. but i found a "solution" to this one. going into lutris and changing the launch option that forces -opengl to -vulkan the store loads up super smooth. but the screen flashes black every 0.1 second. if you're quick enough you can see glimpses of what you want to click and click. as the objects are there, just flashing black like a strobe light. this does this in xorg as well so this problem isn't wayland.

                    in terms of input latency i didn't notice any at all. the mouse moved exactly where i wanted to the moment i moved my mouse. no delay in mouse movement in any of the games i tested. but the classic compositor sluggishness in games is there. i tested in xorg with and without compositor and the outer worlds behaves IDENTICAL in wayland to xorg with compositor on. sluggishness of screen rendering but no mouse delay. but not all games behaved this way. for example wc3 reforged runs as smooth in wayland as it does on xorg with no compositor. stellaris performed no different as well. though that is a native linux game. i tested it with steam runtime and forcing SDL2 for native wayland support. no difference.

                    overall, i really, REALLY like wayland for normal desktop usage. very smooth. firefox on wayland works good. mpv with wayland as well.

                    edit:
                    in xfce with xorg composting on i don't get the epic game store delay but it was sluggish. like playing the outer worlds sluggish with compositor on. but its at least functional. so i don't know whats going on with wayland in that regard since it does it both on gnome wayland and sway.
                    Last edited by pieman; 03-31-2020, 09:15 PM.

                    Comment


                    • Originally posted by sdack View Post
                      Oh, I know. And you're adding more excuses to the pile. Nothing is changing these benchmark results.
                      I play xonotic lot this is first person shooter very latency sensitive. It is a SDL2 game that you can switch between pure wayland and X11 mode.

                      I am not adding excuses to the pile at all. I can tell you that xonotic does more frames per second in wayland mode and you don't have X11 being a frame behind problem.

                      There were no Wayland games in that benchmark but games that support Wayland do exist. . It the legacy games designed for X11 that are slow. Please note there are many legacy Linux games that were designed to use SDL2 that you just update the SDL2 library and they switch over to Wayland backend and pick up the performance boost.

                      Virtual Reality games due to DRM leasing it makes no different what so ever be it X11 or Wayland because the lease bi-passes both.

                      Comment

                      Working...
                      X