A factor of 2 isn't going to be down to kernel differences.
Getting back to the original point. Can you actually cite a situation where the relative slowness of mate matters.
I can also notice gaming performance change when i used to switch DEs/WMs. GNOME 3/KDE are worst of all of course, plain WMs are fine. I think that there is something with composition than interferes with games, try disabling composition in MATE and then try again.
OpenArena is something old and well optimized. The 20% performance difference makes little difference (on most hardware). Say, for example, that I would like to play Tomb Raider 2013. I've already noticed that this game doesn't get as high performance on my hardware on FreeBSD, and I think it uses OpenGL on Wine. It's not that efficient because OpenGL is simply less efficient than Vulkan. Suppose you would achieve 25 fps on average, that is not very playable because you then have dips to, for example, 14 fps and you notice that very strongly. Suppose it would reach 30 fps, that would be considered 'playable'. So it can make the difference between being able to play a game smoothly or not.
If you disable the compositor on an Nvidia card you will have tearing. So that is not an option. Wayland also doesn't work well on Nvidia at the moment.
Ever tried
picom --legacy-backends? It works well for me, while the newer backends are problematic on my 10 years old Intel hardware.
No I haven't tried that. Compton works well the way I set it up but I might give Picom a try with that setting. Is there an advantage to Picom (compared with Compton)?
Your test results are just proofing my argument, which is that these differences don't matter. Why is easy to get: most modern panels use 60 Hz as frame rate, only gamers go and pay for more. So when your panel display a new image every 1/60s, it doesn't matter if a window manager delivers 200 or 300 FPS, because most of these images will never make it to the screen anyway.
Do you think I'm benchmarking Xonotic, OpenArena, and World of Padman because I play this every day?
Well regarding usage of MATE or not that's entirely on your own.
I never said it should matter to you. And there are more important things in life anyway. But you shouldn't make general statements that only concern your situation, that's pretty silly. Here's an example of the kind of statements you've made: Your test results are just proofing my argument, which is that these differences don't matter.
Some Nvidia cards used to be good for mining certain coins. You always saw that many miners did everything to get a few % extra performance from the CPU or GPU.
nl.hardware.info
linustechtips.com
www.overclock.net
With Compton, my GPU usage stays at +- 0% when I use the desktop but do nothing. So it seems that Compton is not causing any real increase in GPU usage.
Because GLX will rely on GPU and will let CPU alone to do other tasks.
That's nice to know.
Yes but keep in mind some games can also use CPU intensively in some situations. I usually disable all kinds of compositions when I run a game like: disable composition (it depends on the DE/WM), run the game, enable the composition back at exit.
If you study my benchmarks carefully you will see that there are tests where MATE gives only 50% of normal performance. MATE has serious optimization problems in some situations.
www.freshports.org
