https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70416
--- Comment #22 from Oleg Endo <olegendo at gcc dot gnu.org> --- (In reply to John Paul Adrian Glaubitz from comment #20) > > Interesting, so it seems the observation you made in the previous comment is > a result of somehow reverting the changes in PR 67391. The changes in PR 67391 are good, but triggered some other problem, i.e. the changes were incomplete. > Looking at the observation in comment #18, it just looks like gcc will > produce slightly larger code. Is this actually a problem which could have > more serious ramifications? As a rule of thumb: more code = more things to execute = slower. That additional 9.1 KByte of code is effectively just a pile of nops which take up time and space.
