Ubuntu 16.10 Desktop Gaming Benchmarks: Unity, GNOME, Xfce, LXDE, KDE, Openbox, MATE

[peppercats]

Were these done full screen or window mode? DEs like Gnome only disable compositing for fullscreen applications, right?

[peppercats]

Now would like to see what those that blame Qt for being "Fat" and "Slow" have to say about those KDE and LxDE numbers.

KDE is fat and slow.

[SaucyJack]

KDE is fat and slow.

On what? A Pentium 4?

[Marc Driftmeyer]

Can't wait until this driver matures:

Gallium 0.4 on AMD POLARIS10 (DRM 3.2.0 / 4.7.0-1-amd64, LLVM 3.8.1)

The Unigine Valley results were:

[dungeon]

Gnome look in line with others composited beside Xfce/Mate and just play CSGO at half speed for some reason

And all that with intel driver, but with other drivers i guess that might be even fine but something else slow... that is how things goes with compositors and gaming

[M@yeulC]

IIRC, there was some talk of a protocol in the KDE/wayland world; that could disable compositing if requested by the application. Maybe Valve games are using it? BTW, the option disable compositor for fullscreen windows has been removed in the latest release.

I would be curious to know if it is possible to disable compositing for only one screen, in a multi-screen setup?

[Guest]

KDE is fat and slow.

As we can see it's fast even with compositions enabled. The same cannot be said about gnome which is buggy. Not only it's slow in CS:GO, but last time I tried it Firefox and Google maps were unusable with nvidia blob. Everything works fine in KDE.

[Marc Driftmeyer]

No need to wait, use amd-staging-4.7 kernel and padoka ppa. Heaven stresses more than Valley and I have this result wit rx460: Unigine Heaven Benchmark 4.0

FPS:	24.2
Score:	609
Min FPS:	6.8
Max FPS:	45.6

System

Platform:	Linux 4.7.0-g1ed6b7d-dirty x86_64
CPU model:	AMD Athlon(tm) X4 845 Quad Core Processor (3491MHz) x4
GPU model:	Unknown GPU (256MB) x1

Settings

Render:	OpenGL
Mode:	1920x1200 2xAA fullscreen
Preset	Custom
Quality	Ultra
Tessellation:	Normal

I tested it and it broke Blender, so I'll have to wait. Could you clarify how come the 8GB for the RX 480 is not cited in the Unigine results, but only 256MB instead?

[andre30correia]

ubuntu is full of slowing bloatware and designed for beginners. I do get why M uses that in gaming tests.

i use ubuntu to develop software and server proposes, is only for beginners? or for all type of ppl? gain some brains, xfce is old, don't work well in new machines, unity and gnome are much better (performance, powersaving, etc)) for gaming without composing or vsync you have a lot of tearing and even in videos but ok, stupid ppl and trolls are stupid and trolls

[Slartifartblast]

Far more comparative if we had Nvidia propriety, AMD propriety and AMD FOSS drivers also compared in the mix, however I understand that there are only so many hours in the day. This is just a test of Intel's shared graphics performance really. I'm going to take these results with a pinch of salt just as the FDA would if this were NDA data and say more testing required, go back to the drawing board.