Ok, so I bit the bullet once more and decided to try KDE 4.3.2 once again on my main machine (Dell GX270, Intel 865 graphics onboard) despite the horrible experience on my AMD Duron (SiS onboard graphics) described in my previous post ‘Goodby KDE4’. To be honest, with not much hope for improvement, as I already had 4.3.1 installed on the very same machine, using kubuntu jaunty - not much better performance there as described above, though.
Anyway, the karmic developers claimed to have improved Intel Graphics support significantly, so I decided to give it one more try. Installation side-by side my production system (Debian Lenny, KDE 3.5.10) went flawless, but slow (as expected). Then boot into karmic, and surprise: this KDE 4.3.2 is no more an awful slug as before ! Not as snappy and responsive as 3.5.10, but really acceptable, finally … (for the first time from any 4.x since I tried 4.0 in January 2008). WoW. So, it seems, KDE4 performance heavily depends on graphics hardware/driver quality, much more than any 3.5.x version before. And I’ve been a victim of my outdated and/or poorly supported graphics equipment, so far.
Appearantly, the new Intel driver in karmic fixed that (almost), at least for my Dell’s. What an odyssee … As a side note, migration of personal settings/preferences (email accounts, kwallet, bookmarks…) went very well, I just copied my ~/.kde folder from Lenny to the new install and everything seems to have been converted smoothly without a hitch, well done, KDE developers! So, yes, I’ll use KDE 4.x again, alternatively to my beloved 3.5.10, see how it improves and eventually make the switch at some point in the future…
Kommentar(e):
deloptes schrieb am 2009-11-29 16:06:05 :
Just read your posting on the debian user list from last week.I spent the hole week trying to install testing with kde 3.5. because kde 4.2 did not provide the expected functionality. I remember when kde 3 came out it was first useful in 3.3, so may be in the 4.3 and above I ll switch to it.I prefer stability over anything else. So why chaniging something that is already working?! Regards