Kevin Kofler wrote: > Solomon Peachy via devel wrote: > > Also, the numbers are purely *GPU* load, ignoring the rest of the system > > (eg the CPUs). Given that Wayland was designed to offload *more* work > > onto the GPU, a higher GPU load in of itself isn't surprising. > > There (from the same author, I am only linking to his posts) you have CPU > and power consumption data: > https://www.dedoimedo.com/computers/plasma-6-4-performance-wayland-x11-power... > > Heck, was any attempt made to repeat things? > > There he did the same kind of benchmarks on a different computer with a > different GPU vendor: > https://www.dedoimedo.com/computers/wayland-vs-x11-performance-nvidia-graphi... > > They used "the same set of programs" but didn't actually enumerate what > > those were (or the workloads). What toolkits/versions? Were any of > > these applications Wayland-native or are they all X11-only? Did any of > > them use some sort of 3D graphics (eg Vulkan or OpenGL)? > > And there he benchmarked some specific workloads: video decoding (at 4K > resolution) and WebGL rendering (the latter uses 3D graphics indeed): > https://www.dedoimedo.com/computers/wayland-vs-x11-performance-amd-graphics.... > And the result is always the same: X11 is faster and consumes less power and > other resources.
Is it? The first article shows that CPU time hasn't really changed and instead just shows a bunch of fluff with the author themselves stating that the perf output cannot be actually compared. The second article shows the same but then that they had a problem getting VLC working correctly followed by data showing Wayland taking more CPU time. How likely is it that this was some kind of issue with the setup instead of an actual difference between Wayland and X11? The third article shows even worse CPU times under AMD while VLC is throwing dynamic linking issues. Besides all of that we don't know: - How exactly tests were ran beyond a vague "it was ran 3-5x" - How long tests were ran - How the data was interpreted to remove statistical outliers - What exact hardware was used during testing - If the tests reproduce outside of the test distro These kind of benchmarks are worthless and cannot be reliably reproduced by anyone. The author has stated that they are going to share more tests using Kubuntu and Fedora however I don't think they are going to start properly testing. Would be nice to see a graph showing how X11 and Wayland behave both idle and under load over a larger time span (say, 24 hours), might be worth trying to do these tests myself. > Kevin Kofler -- _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
