Why does XFCE underpreform in gaming

XFCE seems to promise the opposite, a lightweight DE with little memory usage. Let's go check the charts:
Ouch! XFCE was ranked dead last in all six of the graphs. Why is this? What in XFCE causes this to happen?
 
That was the cause of several discussions on phoronix. In short, and discounting corner cases, XFCE by default enables composition and the wayland DEs used are using direct scan out. If you disable composition the results are more or less similar, some wins for xfce, some for the other DEs.
In doubt, do your tests, if you are a serious FreeBSD gamer it may be beneficial to use KDE/Gnomes with the wayland session, maybe..
 
What performs better if you start a game in X only from a vt terminal? That rules out other causes. Is it xfwm or xfsession? Can't happen anymore.
 
Windows is designed to be a single-user gaming machine.
Under unix[like], so professional and scientific systems, one must always remember, those systems are not primarily designed for gaming, which includes the DEs.
Many games, especially current real time games, are a heavy load to the machine. To run smoothly you not only need powerful hardware match the game's state-of-art, but the system needs to step completely into the background for the game having all hardware resources available, almost running on bare metal. This ain't the default settings on a timesharig multiuser OS where a user always shares the machine with others, no matter how many real users actually using the machine. For realizing satisfying gaming there are several things to change of the system, like ensure no other but only one user uses it, sende several system's tasks to sleep, change priorities, etc.

However,
I quote from the article:

"Both Gnome and KDE were running on the Wayland display protocol, while Xfce was running on X11, and that difference alone led to Xfce falling far behind the Wayland[...]"

So, it's not KDE or Gnome vs Xfce, but Wayland vs X.

This I can comprehend, since X was designed to cover most professional and scientific needs, while Wayland seems to be designed for single-user gaming machines. 😁:cool:
 
vassia the following results done by a third party (some fourth party commentary here: https://hwbusters.com/news/wayland-vs-x11-faster-on-frames-not-on-watts-and-an-xfce-fork-beats-both/) may help you (and others) figure out what is this all about XFCE, composition on, off, wayland and efficiency:

1789553532039.png

(source: https://www.phoronix.com/news/Wayland-X11-Performance-PorteuX )

The following table is also relevant (source: https://github.com/fulalas/xfwm4-gl#xfwm4-gl):

DriverRendererFPSCPU (ms/s)CPU (W)GPU (W)
AMD (Mesa radeonsi)no compositor180.5
-​
18.918.9
XRender161.6
11.2​
21.021.0
OpenGL174.2
13.2​
20.220.2
NVIDIA proprietaryno compositor166.3
-​
4.037.2
XRender160.9
17.5​
5.843.0
OpenGL158.6
16.2​
4.743.2
NVIDIA Mesa (zink)no compositor149.6
-​
6.8N/A
XRender149.1
23.5​
14.8N/A
OpenGL149.1
22.5​
14.8N/A


For reference, XRender is what happens you enable desktop composition on XFCE right now, OpenGL is this team's fork of xfwm to use OpenGL to do composition (not available on FreeBSD).
 
Back
Top