https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63986
--- Comment #7 from Oleg Endo <olegendo at gcc dot gnu.org> --- (In reply to Richard Biener from comment #1) > I think you were mostly lucky And exactly that's why I'd better not rely on combine dreaming up the parallel pattern if I want to control the instruction (sequence) selection in a more reliable way. I didn't mean to file this PR as tree-optimization, but rather as a target issue.
