Ubuntu 21.04 - X.Org vs. Wayland Linux Gaming Performance

[V1tol]

I don't think that results can be significantly faster because obviously it is already limited by resources. But less compositing layers gives lower latency and I suspect those numbers increase thanks to higher minimum FPS. It is even noticeable with Plasma 5.21 on Xorg which has kwin latency adjustment setting. It would be interesting to test one game/app but gather numbers like min/avg/max/95pct etc. - similar to what gaming sites are doing.

[oiaohm]

It's amazing how boring this transition is looking. Once AutoKey is working on Wayland (and I figure out how to disable my broken, ghost-touching touchscreen on a Wayland session), I'm definitely going to attempt the switch.

Define works.

https://github.com/autokey/autokey/issues/87#issuecomment-813569352

AutoKey current version applications running in XWayland on top of Wayland can be controlled. So depending on how you use AutoKey you might be their already.

https://github.com/snyball/Hawck is another interesting one to watch as it can control TTYs as well as graphical.

[treba]

It should be mentioned that unfortunately there's still the issue that if the hardware does not keep up with refresh rate, mouse movement is still throttled to the frame rate - i.e. if the game runs at 40fps on a 60Hz screen, you will notice mouse stuttering. It will hopefully fixed in Gnome 41 - it's not Wayland specific, just a Mutter implementation detail.

That said, I'd expect the biggest performance differences between X11/Xwayland vs native Wayland to show once franctional scaling is properly figured out. On Wayland, it's possible to do it without performance loss.

[gregzeng]

Most other Linux experts remain skeptical about Wayland; Canonical, Red Hat, etc. Fedora is alone with these new untested technologies, including BTRFS.
Wayland needs testing on upscale and downscaling to different display sizes, different & multiple screen sizes, etc. It should be expected that many applications will be very troublesome with Wayland, for several more months.
Both GNOME & Wayland differ from most Linux additions. Only complex modifications can be made before being usable. When Canonical & most others use GNOME, few like the GNOME defaults. Luckily this latest version of GNOME is not such a major deviation from the expected WIMP formula created by Xerox.
Later versions of the official Linux kernels might make these Linux innovations easier. "The Linux Foundation" has so many integration tasks ahead, with Wayland, BTRFS, etc.

[oiaohm]

Most other Linux experts remain skeptical about Wayland; Canonical, Red Hat, etc. Fedora is alone with these new untested technologies, including BTRFS.
Wayland needs testing on upscale and downscaling to different display sizes, different & multiple screen sizes, etc. It should be expected that many applications will be very troublesome with Wayland, for several more months.
Both GNOME & Wayland differ from most Linux additions. Only complex modifications can be made before being usable. When Canonical & most others use GNOME, few like the GNOME defaults. Luckily this latest version of GNOME is not such a major deviation from the expected WIMP formula created by Xerox.
Later versions of the official Linux kernels might make these Linux innovations easier. "The Linux Foundation" has so many integration tasks ahead, with Wayland, BTRFS, etc.

Really the Most other Linux experts. https://wiki.debian.org/Wayland#Supported_environments.

Debian 10 released July 2019 with gnome desktop was defaulting to Wayland over X11 using AMD or Intel graphics. Same is true with RHEL 8 released in 2019.

Sorry to say a lot of Linux experts making distributions have been setting the default to be Wayland progressively more and more.

The writing was on the wall about wayland coming in 2019 as the core distributions that most distributions are based on started defaulting to wayland for their primary desktop environments.

We are almost 2 years in to the conversion over to Wayland over X11.

Wayland needs testing on upscale and downscaling to different display sizes, different & multiple screen sizes, etc.

There is something important here. x.org setup for doing this testing is now only used for testing XWayland and Wayland technologies. Yes Redhat only putting forward a maintainer for XWayland not X11 on bare metal is a big huge change. Going forwards X11 on bare metal having issues with different display sizes is a real risk due to no maintainer and no testing framework to detect those faults.

[HarlemSquirrel]

It would be cool to see some Valheim benchmarks

[lyamc]

Personally I’m waiting for a VNC or VNC alternative before I make the switch to Wayland

Edit: I am aware of wayvnc, it does not currently work with gnome

[leipero]

I like how, aside from the one game, the Wayland performance seemed better if only by a slight hair. To me, minimal FPS and and frame rate are the more important numbers so it's nice seeing XWayland having slightly better numbers even if it isn't by much.

Since I've been gaming on Xorg KDE and Windows lately -- Can anyone comment on how input feels on Wayland? I assume it's nice and snappy. PS4 BT Input lag is my biggest worry when the Wayland transition finally happens on my system.

It's not there yet. It's much better on GNOME 40, but if using WINE for example (only tested on WINE with one game to be fair) input latency is still an issue, much less so than what it was in GNOME 3.38.x but it is still there. Now it's almost acceptable, lower latency than X11 with VSYNC, but worse than X11 without VSYNC.

[jacob]

Really the Most other Linux experts. https://wiki.debian.org/Wayland#Supported_environments.

Debian 10 released July 2019 with gnome desktop was defaulting to Wayland over X11 using AMD or Intel graphics. Same is true with RHEL 8 released in 2019.

Sorry to say a lot of Linux experts making distributions have been setting the default to be Wayland progressively more and more.

The writing was on the wall about wayland coming in 2019 as the core distributions that most distributions are based on started defaulting to wayland for their primary desktop environments.

We are almost 2 years in to the conversion over to Wayland over X11.



There is something important here. x.org setup for doing this testing is now only used for testing XWayland and Wayland technologies. Yes Redhat only putting forward a maintainer for XWayland not X11 on bare metal is a big huge change. Going forwards X11 on bare metal having issues with different display sizes is a real risk due to no maintainer and no testing framework to detect those faults.

X11 on metal should be pretty much considered abandonware at this stage. But IMHO RedHat should announce an EOL date even for XWayland. It can still be far in the future, say 2025, but there should be a clear message to third parties that they must prepare for the switch and not just stick to X11 virtually indefinitely. This is especially the case for Steam and Steam games.

[arun54321]

Yes Yes. Die Xorg trash.