http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48636
--- Comment #20 from Jan Hubicka <hubicka at gcc dot gnu.org> 2012-10-16 16:38:27 UTC --- Created attachment 28456 --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=28456 Path I am considering Hi, I am considering to enable inlining when inline-analysis says that the inline function will get significantly fater regardless the inline-insns-single/inline-sinsns-auto limits. The patch is attached. In general it may make more sense to gradually set the limits based on expected speedup but I am affraid this will become hard to understand and maintain. So for now lets have simple boolean decisions. It would be nice to know where it helps (i.e. for fatigue and cray) and where it doesn't. It also causes quite considerable code size growth on some of SPEC2000 for relatively little benefit, so I guess it will need more evaulation and reduction of inline-insns-auto limits. It also may be problem in unrealistic estimates in ipa-inline-analysis, this is first time we take them really seriously. Comments/ideas are welcome.