Re: GTK is double buffered by default ? on GNU/Linux and MS Windows ?



Hi again,

Feel free to work on this and submit patches.
Well I am not interrested in the win32 port. I did already some
experiments with tuning GTK on Linux which resulted in a 25%-35%
performance increase on nvidia hardware.
I also talked with nvidia driver developers about existing problems
... and how they eventually could solve it in their drivers.

Maybe you should use one of them instead, then?
Yes right - I try to do this whenever possible. I ran Eclipse hooked
on top of Fox as long as SwtFox worked with the Eclipse versions I
use, I use KDE and all of its tools.
But sometimes I simply can't get arround GTK like when using FireFox,
Thunderbird, gaim, gimp or when using my Nokia770 - which is quite
fast but definitivly underpowered with its 250mhz arm9 to resize two
lists fluently (fltk had no problems at all on the same device,
gtk-resize 1fps, fltk:8fps). PalmeSource and Nokia stick at gtk-2.6
because 2.8+ is simply to slow for embedded use.

I just think its funny how GTK devs do proceed with preformance
related posts. First people are told that the problem they are
reporting are very vague and that this is no proof that GTK is slow.
Then 50% of the compaining people are quiet. The other 50% submit
performance data and results - since these guys can proove that its
slow they don't get any replies any more - since they can't be blamed
for flaming and nobody feels responsible for investigation.

This makes me a bit sad when thinking about the fact that almost
everybody in the OSS world is moving to GTK. When looking at blogs
like this: http://www.sysinternals.com/blog/2006/07/first-week.html
Ms is investigating any report with experts analysing problems through
the whole chain of involved components. People report that they have
problems and get fooled and ignored ... well .. thats the other way to
do it. I wonder who's the fool ... the people complaining or the
maintainers ignoring 1+ performance complain per week.

Best wishes, lg Clemens



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]