Ref http://www.linuxfromscratch.org/lfs/view/development/chapter06/gcc.html
Just an observation, but compiling gcc-4.6.1 like this, i.e. with "-fomit-frame-pointer" results in a gcc which produces compiled output +/-15% larger (after stripping) than using gcc compiled without "-fomit-frame-pointer" when compiling with CFLAGS="-march=i486 -mtune=i686 -Os -pipe" Note that (per gcc docs) as from gcc-4.6.x, using the optimization -Ox (i.e. -Os, -O2, etc) automatically causes "-fomit-frame-pointer" to be used and the ./configure switch "--enable-frame-pointer" is required to reverse this. I don't know if "-fomit-frame-pointer" produces compiled output that is more efficient (i.e. runs more quickly), but a +/-15% size increase is significant - even the gcc compiled by a gcc without "-fomit-frame-pointer" will be significantly smaller... Regards John -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
