Lunar optimize

[Jean-Michel Bruenn]

Hello,

the thing with that is, that "native" is THE safest optimization for
your CPU. However, if you're compiling for another box or for a
chrooted environment (where native will break the whole chrooted
environment) you're an experienced user who knows what he/she/it is
doing. So just change safe to "no" (so that you can see the unsafe
optimizations) and switch the cpu optimization to whatever you like
and don't touch anything else there.

Cheers
Jean

On Wed, 29 Sep 2010 14:21:23 +0200
"Guillaume" <sirocco at gmx.fr> wrote:

> > On 09/28/2010 02:59 PM, Guillaume wrote:
> >> Hello,
> >>
> >> It seems lunar changed the way we can define the CPU with GCC 4.5 in "lunar optimize".
> >>
> >> I don't want to use native on my core2 because I sometime create modules for my slow atom
> >> computer.
> >> With GCC 4.4 I can select any architecture I want :)
> >> With GCC 4.5 I only "native" is available  :(  (I don't know architecture "None")
> >>
> >> Is there any plan to add more architecture to GCC 4.5 settings ?
> >> If not I'll manually modify /etc/lunar/local/optimizations.GCC_4_5, could you please confirm
> >> that updates of lunar module will not overwrite this file.
> >
> > Native is correct as base optimization for your atom processor. That's
> > what you should use.
> >
> > Auke
> >
> 
> I agree that native is the correct optimization for any computer that compile modules for itself.
> But my atom also use applications compiled by another PC ... so I can't set native the other PC.
> This may looks strange to you but it save me a lot of time :)
> 
> I just hope that CPU=native will not be hard-coded in lunar code but will remain in
> optimizations.GCC_4_5 file.
> 
> Drew is true ... I should have a look at distcc, I just don't know how complicated it is (this is
> why up to know I share some compiled modules between computers).
> 
> Guillaume
> 
> _______________________________________________
> Lunar mailing list
> Lunar at lunar-linux.org
> http://foo-projects.org/mailman/listinfo/lunar


-- 
Jean-Michel Bruenn <jean.bruenn at ip-minds.de>