GNOME 3.26: Wayland vs. X.Org Performance - Boot Times, Power Use, Memory Use & Gaming

Written by Michael Larabel in Software on 11 September 2017 at 11:18 AM EDT. Page 1 of 3. 98 Comments.

While testing out the near-final GNOME 3.26 this weekend I also ran some benchmarks of it comparing the boot time, memory use, power consumption, and gaming performance when comparing GNOME Shell / Mutter running on Wayland and then an X.Org session.

From the Razer Blade Stealth laptop with Intel Core i7 7500U Kabylake processor and HD Graphics 620, I ran some quick tests of the near-final GNOME 3.26 packages. Tests were done using the Fedora 27 development repository with the Linux 4.13 kernel and Mesa 17.2.

The same tests were done when logging into an X.Org session and then again under the Wayland session. All tests were done via the Phoronix Test Suite.

First up is a look at the load time via the reported systemd boot times when looking at the user-space timing difference for the boot process to each session with GDM being set to auto-login to the respective account.

The user-space start-up time was quicker with the Wayland-based GNOME 3.26 Shell session taking just 22.4 seconds compared to 25.3 seconds reported by systemd when using the GNOME Shell on X.Org.

The memory use was looking at the freshly booted system's memory usage from Fedora 27 Workstation under X.Org and Wayland while idling at the desktop and no other applications/windows open. The difference wasn't huge, but interestingly the memory usage under Wayland was slightly higher. Perhaps it would be lower when Mutter is able to drop its various X11/XWayland dependencies when not needed.

The battery power use of the Razer Blade Stealth while idling within GNOME on X11 and Wayland was similar with only small differences discovered.

Next onto some OpenGL benchmarks.


Related Articles