Announcement

Collapse
No announcement yet.

Ubuntu 21.10 Radeon Gaming With KDE Plasma vs. GNOME Shell + Wayland vs. X.Org

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

  • #41
    Originally posted by mirmirmir View Post

    I pick kitchen knife for kitchen stuff 10/10
    And if computers were vegetable this analogy may have been applicable. The truth is that Plasma 5 is very usable DE with active developers and its default appearance and behavior is very close to Win10. It also has UI modules and modes for mobile devices with touch screen and the likes. Valve made the right call here.

    Comment


    • #42
      Originally posted by Danny3 View Post

      And if you want desktop icons or minimize button in Gnome 3, you do what ?
      You need to use workarounds and hacks like third party extensions !
      That's not the "right tool for the job" point that you want to make.
      I don't, i don't need desktop icons, i run my app on app launcher, i keep my files tidy on their designated folder. I don't need minimze as i use workspace. Scrying on small bottons on the bottom of the screen is tiring. The idea of minimizing apps you're using is kind of stupid. Why don't the desktop do the job and bring my window right in front of my eye?

      Comment


      • #43
        Got some fresh benchmarks for y'all !!

        Wayland is legit. It was actually slightly faster than Xorg in my benchmarks.

        Benchmarks:

        pts/osbench
        pts/ctx-clock

        Too lazy; didn't click:

        osbench:

        Code:
        Geometric Mean Of All Test Results
        Result Composite - osbench-21-10
        Geometric Mean > Higher Is Better
        21.10 fresh install, mitigations=on . 3.295 |================================================= =
        mitigations=off, preload ............ 3.761 |================================================= ========
        mitigations=off ..................... 3.771 |================================================= ========
        5.15-rc6 custom kernel, Xorg ........ 4.692 |================================================= ======================
        5.15-rc6 custom kernel, Wayland ..... 4.708 |================================================= ======================
        Do you want to view the results in your web browser (Y/n): n
        Would you like to upload the results to OpenBenchmarking.org (y/n): y
        Would you like to attach the system logs (lspci, dmesg, lsusb, etc) to the test result (y/n): y
        ctx-clock
        Code:
        Clocks < Lower Is Better
        21.10 fresh install, mitigations=on . 1015 |================================================= =======================
        mitigations=off, preload ............ 181 |=============
        mitigations=off ..................... 181 |=============
        5.15-rc6 custom kernel, Wayland ..... 154 |===========
        5.15-rc6 custom kernel, Xorg ........ 154 |===========






        It's honestly crazy how fast my custom kernel is (yeah, I'm bragging: deal_with_it-shades.gif). (edit: pronounced jif)

        So while out-of-the-box Ubuntu is lightening fast, y'all are leaving *tons* of performance on the table still. I'm just saying, look into it.

        (PS: sorry to blow up the thread with screeenshots. But hey, we want The Need For Speed, right? That's why we're all here, right? Just checking.)
        Last edited by perpetually high; 19 October 2021, 05:16 PM.

        Comment


        • #44
          Originally posted by perpetually high View Post
          Got some fresh benchmarks for y'all !!

          Wayland is legit. It was actually slightly faster than Xorg in my benchmarks.
          I think this is a joke, right? I can't quite tell.

          Comment


          • #45
            Originally posted by smitty3268 View Post

            I think this is a joke, right? I can't quite tell.
            Where’s the joke? I ran pts/osbench. Are you suggesting there would be no difference between the two for those benchmarks? I was curious for my own sake so I ran on both. Let me know where the joke was. I want to laugh.

            Comment


            • #46
              Originally posted by schmidtbag View Post
              So really, the takeaway here is Wayland's overhead is astonishingly minimal.
              Not "really" at all, but yeah: the important piece is that the overhead isn't important enough to actually matter.

              wrt to some of the other posts: nobody sane is going to target Wayland directly. You can echo-chamber the hype all you want, but it's still only on a fraction of systems, and that's not going to change until either Wayland gets a lot closer to parity with X - which is something that we know is never going to happen - or X *over* Wayland makes it happen transparently-ish, such that everyone is "really" running Wayland, but programs are still using the X APIs.

              Engines will add Wayland backends over time, and some probably already have them at this point, but it'll be years at best (and possibly "never") before Wayland becomes a worthwhile primary target. So it's great that XWayland is working well, because that's the part that DOES matter ATM, and will continue to be for a very long time yet. Wayland is "the future" only in the sense of "XWayland rather than X directly". That's still progress though, so it's great that XWayland is working so well.

              Comment


              • #47
                Originally posted by perpetually high View Post

                Where’s the joke? I ran pts/osbench. Are you suggesting there would be no difference between the two for those benchmarks? I was curious for my own sake so I ran on both. Let me know where the joke was. I want to laugh.
                Are you claiming that the graphical environment is affecting the speed that the kernel can create files? By making it 1% faster?

                No. That's simply standard margin of error.

                Comment


                • #48
                  Originally posted by smitty3268 View Post

                  Are you claiming that the graphical environment is affecting the speed that the kernel can create files? By making it 1% faster?

                  No. That's simply standard margin of error.
                  I've ran these tests before. It was more than margin of error. I can almost predict these scores down to the decimal point.

                  I get that I didn't run graphical tests, but Wayland is a separate binary, it's worth it to look at how it affects overall OS benchmarks. But your point is not lost on me, that's why I provided benchmarks (but it's more than just the margin of error).

                  EDIT: smitty, keep also in mind that PTS runs each test three times, then averages that score, so it's not just running the test one time. You may know that already, but sharing it for others. That helps eliminate margin of error. I've ran these tests a bunch of times. I can predict how it'll reach after a certain tweak. Pretty cool actually. When you remove all the noise, things should actually be pretty predictable.

                  Another two runs with Xorg vs Wayland with some various env variables / C FLAGS (and re-compiling the kernel with said flags (see here if curious: https://pastebin.com/ApKcXTf5)

                  https://openbenchmarking.org/result/...TJ-OSBENCH2174

                  Anyways, it's pretty predictable. Some tests Xorg is faster, some Wayland is faster. But consistent on which is which. Yes, at this point, it's getting around margin of error territory, but still predictable. Let me know if you vehemently disagree but I feel good about it.
                  Last edited by perpetually high; 20 October 2021, 07:35 AM.

                  Comment


                  • #49
                    smitty3268 ran some more tests, Wayland is definitely faster

                    https://openbenchmarking.org/result/...09&sgm=1&ppt=D

                    (Note: mitigations=off intel_pstate=disable acpi-cpufreq performance was for all tets, not just the first. Didn't bother with the "default slow" benchmarks, but they were considerably slower on a fresh install and mitigations=on for Haswell, fyi)

                    Memory Allocations was noticeably faster:



                    Last edited by perpetually high; 26 October 2021, 08:57 PM.

                    Comment

                    Working...
                    X