Gaming/Graphics Performance On Unity, GNOME, KDE, Xfce

[kraftman]

They already rooted out some excessive and obsolete stuff like Kubuntu.

You could keep for yourself such bull. Kubuntu was Ubuntu killer, so they had to drop it. If Kubuntu was obsolete then everything was.

[Julius]

You need to test this with a more modern game... maybe Oilrush or Xonotic would be an option? As it is these benchmarks hardly reflect real life gaming performance unless you play these games with old tech.

[GreatEmerald]

Yea, KWin doesn't unredirect fullscreen windows at all by default, because it causes annoying screen corruption when combined with overlay windows (like the Amarok OSD). So while it should be considered a bug, it still explains the poor performance. To be fair, all tested games should have been run in windowed mode, in order to truly compare the compositing performance, and not the unredirection performance.

Though at least for me, using latest Catalyst drivers with KWin causes screen blackening, so I have to keep it off regardless...

[xpander]

you forgot the MATE desktop.
but other than that quite expected results

[drago01]

Yea, KWin doesn't unredirect fullscreen windows at all by default, because it causes annoying screen corruption when combined with overlay windows (like the Amarok OSD). So while it should be considered a bug, it still explains the poor performance. To be fair, all tested games should have been run in windowed mode, in order to truly compare the compositing performance, and not the unredirection performance.

Though at least for me, using latest Catalyst drivers with KWin causes screen blackening, so I have to keep it off regardless...

http://blog.fishsoup.net/2011/06/13/benchmarking-compositor-performance/

[gedgon]

Unless the WM is doing something stupid, all of the non-compositing ones should have about the same performance.

Indeed. I have no idea how Michael gets these results for uncomposited DE.

Here's lightsmark on GTX 275/C2D E4500

[GreatEmerald]

http://blog.fishsoup.net/2011/06/13/...r-performance/

Thanks, that was a very interesting read!

[energyman]

in a perfect world the window manager should have ZERO impact on gaming performance with a fullscreen 3d game.

Which makes all results with the nvidia driver look VERY fishy. Something is not done right there.

[korrode]

i was wondering the same thing. personally i like lxde (i'm using it right now) and i found it to be noticeably faster than xfce. i personally hate xfce, i'd rather use unity. back before the gnome 3 days, xfce was just about as heavy as gnome 2, but offered less features and yet it strived to be lightweight.

I don't know what distro you use or how you obtained your Xfce setup, but Xfce's performance is regularly ruined by changes/additions made by distributions (like the mountain of sluggish fail that is Xubuntu).

I ran openbox for years, then i ran fluxbox for years. I'm extremely minimalist and greatly value responsiveness from my desktop.

These days I run Xfce, session managed and all, on ArchLinux and am happy with how it performs.

How you can be a fan and user of LXDE but say something like "i'd rather use unity [than Xfce]" boggles my mind. But each to their own.


<insert fanboyism here over Xfce trumping the competition in the benchmark results>

[Bestia]

Oil Rush performance

I've tested performance of Oil Rush on my PC running Ubuntu 11.10 with different shells:

AMD Athlon(tm) 64 X2 Dual Core Processor 4400+
nVidia 9600 GT NVIDIA Driver Version:280.13
2 GB RAM

fulscreen 1920x1080, anti-aliasing off, graphics medium, textures high

Enlightment E17 (with ecomporh enabled) from 39 to 60 fps (mostly 46-50 fps)
Razor-qt from 37 to 60 fps (mostly 41-50 fps)

Unity (using Daniel van Vugt PPA's for Compiz and Unity)
Unity 2D
Gnome Shell
Gnome Classic (no effects)

all four of them are locked at 30 fps (sometimes going to 28 fps)

I've used built-in fps counter of the game and played on the same map always loading the same save.