https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125809
--- Comment #4 from Drea Pinski <pinskia at gcc dot gnu.org> --- (In reply to ptomsich from comment #3) > The .text section increases by ~1MB, so I expect issues on the > frontend/i-cache side. Note text section increasing does not seem like corresponding to the 8% slower either. There has to be something else. Maybe a later on optimization after the inlining/const prop has happened that causes the slow down. Like for an example a vectorization that was not happening before and the extra branches to the scalar code is causing the slow down. There is a lot of things that happen after clonining which could cause problems. But just saying the cloning is not profitable is not correct here.
