I’d be interested in tests across hardware of different ages (but as others have said, same/similar DEs/WMs/compositors as much as possible, please). While I have no proof, I’d guess that Wayland is making assumptions that would make it less efficient on some older kit.
For dual monitors and partial scaling wayland have been a bless.
It’s not that’s more or less efficient or more of less fast. It’s that’s been able to do things thar X was unable to and had to be practically hacked to get a clunky solution.
It’s better. Much better. If Linux desktops was still on X, it would feel and be incredibly dated. I know it works “fine” but has many limitations, which is why we needed Wayland in the first place.
Didn’t because Wayland can produced more performance and that’s why it’s using more power?
Only ran once, doesn’t measure x server cpu usage, claims “not necessarily more efficient” despite kwin and labwc showing higher framerate for seemingly similar power usage numbers, no discussion of e.g. frame times, consistency and cpu vs gpu limited…
I like Wayland just because it lets me easily use two monitors of different resolutions and refresh rates.
I used X11 for about 6 months when I first used Linux because I still had an Nvidia GTX 1080 at the time and Wayland wouldn’t even load for me then.
I did manage to get my monitor setup somewhat working on X11 through a lot of manual editing and the trade off constant screen tearing on my low hz 2nd monitor.
When I finally got an AMD card I loaded KDEs Wayland session on the same Linux install and it automatically set the max res and hz for both monitors including the new 1440p 170hz main monitor I got along with the GPU.
Doesn’t KDE kinda have that same refresh rate issue on Wayland? I remember seeing a bug that Qt has a limitation in this regard, In the video I saw the workspace transition ran at the lower refresh rate when multiple monitors were used.
Thankfully I stopped using multiple monitors so I can’t personally confirm.
KDE plasma wayland user since 5.20 specifically because the x11 version would bug out on me having a 240hz monitor alongside my 60hz drawing tablet. Plasma 6.7 has fixed basically every bug with refresh rate/animations I’ve had at this point including that frame rate one(which I think was caused by the animation reading from disk instead of ram?)
Wayland is just the protocol. You’re benchmarking a bunch of different wayland compositors against XOrg, and from the looks of things, at least one (mutter) is worse in every metric than XOrg, so the headline of the article is misleading.
Phoronix’s more real world testing didn’t have Gnome performing badly: https://www.phoronix.com/review/cachyos-desktops-july-2026
Mutter is made by GNOME, what do you expect?
They compare different desktops, ok. But what about XFCE X11 vs. XFCE wayland?
In addition to that, it would be nice to see comparisons including xlibre, since it is slowly being adopted by certain distros and bsds.
This is stupid and probably ignorant too, but I couldn’t get RDP dynamic window resizing to work on Wayland no matter what I tried and went back to gnome which works great.
Gnome isn’t orthogonal to Wayland.
You’d need to check if you were using the X11 or Wayland session of Gnome
Ok… ? Then it was X11 that worked not Wayland… ffs
Gnome has defaulted to Wayland for a long time, so it sounded more like you had an issue with a different desktop environment, not X11 vs Wayland
I’ve been running nothing than sway after doing a quick test on my rpi 4, running doom 3 – and the performance difference was really significant.
