KDE Versus Xfce

[Auke Kok]

Brian J. Tarricone wrote:

>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA1
>
>On 12/20/2005 3:45 PM, Erik Harrison wrote:
>  
>
>>Lubos Ludak has been doing some performance profiling for KDE. He's
>>been using Xfce startup as his benchmark. Given dropping some things
>>from KDE to "match" Xfce featurewise, he's matched us for startup.
>>
>>He also nicely provides the bootchart, showing a profile of Xfce's
>>disk and CPU utilization. He even capitalizes Xfce right.
>>
>>http://www.kdedevelopers.org/blog/280
>>    
>>
>
>Ugh, xfdesktop takes forever to start, and I imagine it's worse in SVN
>(well, actually, maybe a draw, since I improved the initial menu
>generation time).  Methinks I need to sit down with sysprof for a while.
>
>Would be nice if xfce4-session could start stuff in parallel, though I
>figure that's probably non-trivial to implement, else Benny would have
>done it.  That chart seems to show some sleeping slices (disk I/O?) near
>the start of all apps such that parallel startup might yield faster
>overall startup.
>


it's really too bad that this is the case - but when I see the stats I 
wonder about other things like memory usage - possibly KDE is much more 
optimized than Xfce - after all they have a tonload of good coders doing 
nothing but profiling, so you would expect that they are able to squeeze 
everything as fast as possible into the system. Xfce is probably smaller 
but wastes time on the fringes everywhere - the performance gain 
achievable for Xfce should be much higher than for KDE and GNOME - 
they're at the end of their possibilities.

Auke