On Wed, Jun 07, 2006 at 07:42:36PM +0200, Anselm R. Garbe wrote:
On Tue, Jun 06, 2006 at 11:36:43AM +0100, Simon Morgan wrote:
On Tue, Jun 06, 2006 at 02:38:00AM -0400, Kris Maglione wrote:
> It's late here, so let me clarify that I sometimes forget that people on
> mailing lists use anything other than the development head of a given
> program. The wmii releases specify -O2 (although -O3 works perfectly well
> and gives noticable speed improvements) and do not include debuging
> information. The binaries, however, are still not stripped. strip -s
> removes ~15k, or about 10% (Note that a stripped Ion is ~328k, fluxbox is
> ~8.4M).
wmii-3 does't seem to use -O at all, unless I'm missing something.
Yes that is true. I think -O is no good default. The default of
the compiler should be the safest method, otherwise the compiler
is broken. If someone needs -O666 he can use this option on his
own.
(I really doubt that you notice any difference)...
Actually, I've done some testing and there really are significant performance
increases in some areas, even in -O3 over -O2 (O3 does inlining), especially
in the areas of window moving and dragging (these are *much* smoother with
-O?). Most other operations seem much snappier too. I haven't tested my load
w/ and w/o -O, but I would imagine it would be somewhat lower with it.
--
Kris Maglione
If not controlled, work will flow to the competent
man until he submerges.
_______________________________________________
[email protected] mailing list
http://wmii.de/cgi-bin/mailman/listinfo/wmii
