https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94960
--- Comment #15 from Jan Hubicka <hubicka at ucw dot cz> --- > Oh, sorry, that was linked earlier. But still, isn't the problem that "inline" > is too strong? Do we have some data on this? I plan to do some inliner benchmarking over christmas like I do every year. With cumlated mod-ref, ipa-prop and early opts improvements it may let me to reduce difference between agressive (declared inline) and conservative settings. But nothing dramatic. Overall libc++ seems to be more happy to delcare relatively complex things (like push_back reallocations) as inline+always_inline and clang is more agressive about inlining at -O2, so most of the PRs we have where clang wins is simply due to fact that it inlines everything. I included some size compares in some of the patches and they typically end up with larger code. On the other hand our -O3 inlines some bigger functions clang does not and we are also much more agressive on inlining cross-modularly with LTO.
