Are Compositing Window Managers Lightweight?

[gedgon]

Bullshit. The simply action of moving windows takes up next to zero CPU time with GL compositing. Without compositing it can be very CPU consuming, esp. when the window is big.
Anybody can just open a CPU monitor and check for himself.

Like I said here http://phoronix.com/forums/showthrea...ht=#post318795


OGL KWin
6-8% - Kwin
2-4% - X

Openbox
4% - X
2% - Openbox

Sorry for the horrible quality

[kokoko3k]

Assuming that your system is "balanced" between cpu and gpu, compositing is a must; at least this is what i see on a poor atom n270+gma945

[kokoko3k]

Like I said here http://phoronix.com/forums/showthrea...ht=#post318795


OGL KWin
6-8% - Kwin
2-4% - X

Openbox
4% - X
2% - Openbox

Sorry for the horrible quality

i'd try the same test by stacking something more, you jusy have 2 windows and nothing in the root one.
Maybe i'm wrong, but i think that focing X to repaint at least a background image (read - if not plasma-desktop, get a wallpaper, at least) could change your observations.

[GreatEmerald]

Have you tried yet? Go ahead and try it!!!!

You couldn't be more wrong.

Interesting. Moving a KWrite window around very rapidly with OpenGL results in KWin taking 5-6% of processor usage, and Xorg takes 3%. When compositing is disabled, KWin does not use CPU, Xorg uses 9%. Overall it makes no difference, except for eye-candy. Though it makes me wonder what I'd get with all effects disabled - I just don't want to test, as I have a plenty of effects and I'll forget which ones I had on beforehand...

[gedgon]

i'd try the same test by stacking something more, you jusy have 2 windows and nothing in the root one.
Maybe i'm wrong, but i think that focing X to repaint at least a background image (read - if not plasma-desktop, get a wallpaper, at least) could change your observations.

wallpaper + 10x kate window all over the desktop + konsole

KWin - ~17,5%
X like before

Openbox - like before
X - ~15%

[Awesomeness]

Have you tried yet? Go ahead and try it!!!!

You couldn't be more wrong.

Why are you asking such stupid questions? Of course I already tried it and I know for a fact that I'm right.

[Awesomeness]

Interesting. Moving a KWrite window around very rapidly with OpenGL results in KWin taking 5-6% of processor usage, and Xorg takes 3%. When compositing is disabled, KWin does not use CPU, Xorg uses 9%. Overall it makes no difference, except for eye-candy. Though it makes me wonder what I'd get with all effects disabled - I just don't want to test, as I have a plenty of effects and I'll forget which ones I had on beforehand...

Obviously CPU utilization is not set in stone since we all do not have exactly the same hardware. However the difference becomes clearer the larger the moving window is and when the window movement is diagonally.

On my now almost 5 years old computer without any compositing moving a window which is almost full-screen causes the KWin and X processes to consume ~25% each. With compositing X idles at 2 or so percent and KWin uses 5?7% ? no matter which effects are enabled.
Not only that but the window feels more fluent, especially when moving the window diagonally. When Xorg has to repaint parts of the screen on all sides of the window, the repaint rate drops noticeably.

[Ericg]

It does decrease CPU and power usage. My netbook's battery life is better with compositing enabled on Xubuntu 12.10.

Correct. CPU can DO the same things the GPU can do it, it just takes longer. Longer time means more spent wake, means more time hogging the CPU (bad if youve got other things happening or trying to), means more power usage. Martin did a blog post about why Compositing is better enabled for battery life vs not enabled.

Though, I will say 2 things.... 1) Martin's reasoning does make sense. But why does Windows 7 disable compositing still under powersaver mode? I'd think Microsoft of all people would know whether or not compositing positively or negatively affected battery life.

2) Why does PowerTop report better battery life without compositing than with?

[GreatEmerald]

Obviously CPU utilization is not set in stone since we all do not have exactly the same hardware. However the difference becomes clearer the larger the moving window is and when the window movement is diagonally.

On my now almost 5 years old computer without any compositing moving a window which is almost full-screen causes the KWin and X processes to consume ~25% each. With compositing X idles at 2 or so percent and KWin uses 5?7% ? no matter which effects are enabled.
Not only that but the window feels more fluent, especially when moving the window diagonally. When Xorg has to repaint parts of the screen on all sides of the window, the repaint rate drops noticeably.

Hmm, if moving an almost-fullscreen window, with compositing I get kwin at 7% and Xorg at 3%. Without compositing it's kwin at 3% and Xorg at 12%.

[Awesomeness]

Hmm, if moving an almost-fullscreen window, with compositing I get kwin at 7% and Xorg at 3%. Without compositing it's kwin at 3% and Xorg at 12%.

As I wrote: My PC in ancient, so the difference is more drastic.

The ones who advertise that GL compositing is bad, probably own relatively new PCs with many CPU cores and only think they know what's better for lower-end hardware. The fact is that the older a PC is, the more help its CPU needs from the GPU.