E17 benchmarks

samuel verstraete samuel.verstraete at gmail.com
Thu Jun 9 11:02:26 CEST 2005


Just did some testing myself with this benchmark...

The first one just checks how many windows users can draw in one second. For that it dumps a whol lot of windows on the desktop one after each other. Xfwm4 handles this pretty well just untill we reach a certain treshold and then the performance just collapses... (dramatically) i think after the treshold we only have 1 new window every 2 seconds... I don't know how relevant this test is... As Xfwm4 handles the first +100 windows relatively fast i wonder how bad it is... I would like somebody to look into that c code and change the amount of windows that you he will dump...

The second one tries to draw a lot of windows as well but here we are closing them every time and we check how long it take to draw a new window...  As such we are measuring responsetime... imho this is a much more important benchmark as responsetime will determine to a large amount the user experience... It's a little hard to get a good view on this but i have the impression that 4.2.x scores better than the 'old' 4.0.x (the one rasterman uses)... Still i'm only talking about percentages... not talking about faster than gnome f.e. 

One further remark... Memory usage on both of these tests stayed very low with xfce4 (didn't test any other wm's yet)  but cpu usage just tops out... 

Right now i'm looking how it behaves if i let it draw less windows

Samuel

On Thu, 9 Jun 2005 09:54:53 +0200
Jasper Huijsmans <jasper at xfce.org> wrote:

> Hi Olivier,
> 
> Just saw some benchmarks by rasterman for E17. Xfwm4 comes out at the bottom,
> far below metacity. That sounds wrong to me. Maybe it's useful, maybe not...
> 
> http://www.rasterman.com
> 
> 	Jasper
> _______________________________________________
> Xfce4-dev mailing list
> Xfce4-dev at xfce.org
> http://foo-projects.org/mailman/listinfo/xfce4-dev



More information about the Xfce4-dev mailing list