Post by @withaveeay@mastodon.scot

2026-02-25 09:18 UTC

@lproven @fedops Without wishing to fan the flames of this discussion, there's one oddity about Xorg vs Wayland on my laptop, running Plasma. Wayland uses about 1w less power than Xorg on openSUSE, at least. I can't run the test definitively under my preferred MATE, as it doesn't support Wayland, but its at-rest power is roughly the same as Plasma under Xorg. It gives a substantial increase in use on a single charge. I don't understand why this should be the case, but it is.

Replies (2)

@lproven@social.vivaldi.net 2026-02-25 14:41

@withaveeay @fedops I don't know. I have discovered something interesting recently that may be tangentially related. I have 2 old Thinkpads with onboard Nvidia GPUs I can't upgrade. They need an old Nvidia proprietary driver that doesn't work on kerne 6.5 or later. No modern distro using X.org supports the secondary GPU. Many can't wake the Displayport at all; some can, but rendering is by the on-chip Intel GPU. So, they mainly run Ubuntu 22.04 as its original kernel supports driver v390 fine. But 2 modern Wayland distros can use it. Ubuntu 25.10 with GNOME, and Pop OS 24.04 with COSMIC. Questing's GNOME _only_ supports Wayland; COSMIC is Wayland-only. But recent Wayland compositors have some built-in equivalent of Nouveau that can see, identify, and activate the 2nd GPU. Suddenly they work with newer kernels, but not via any X.org desktop. No Unity, no Xfce.

Open ##1419198
@withaveeay@mastodon.scot 2026-03-01 14:24

@lproven Just FYI, classic sample-of-1 problem. I discovered that the difference in power use between xorg vs wayland in wayland's favour was caused by the fact that I'd set my monitor to a 60Hz refresh rate on wayland exactly to save power, while I'd forgotten to do that under xorg. There's still a difference, but negligible given the test conditions. So this killer reason to use wayland is not the case at all.

Open ##1419199