On 2010-06-28 21:32 +0200, Sven Joachim wrote: > On 2010-06-28 21:07 +0200, Tomasz Maluszycki wrote:
> > (for me -O3 is stable... I didn't had any problem with it) > > Depends on the program, I think. In general, -O3 produces bigger > binaries (than -O2) that may or may not run faster. Mozilla even > builds with -Os by default. You'd be surprised at the optimisations that GCC misses on i386. I've seen it needlessly reload registers on every iteration on a tight loop. Neither -O2 nor -O3 changed that. The only way to make it generate sane code was -Os. These days, it's not safe to assume that -O3 produces the fastest binaries. Always benchmark it against -Os. -- André Majorel <http://www.teaser.fr/~amajorel/> bugs.debian.org, your one-stop shop for email addresses. -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100629071142.ga14...@aym.net2.nerim.net
