https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84106
Richard Biener <rguenth at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |rguenth at gcc dot gnu.org --- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> --- Try -fno-tree-loop-distribute-patterns. The test2 loop is recognized as memcpy and expanded as such. Theory (and practice?) is that the emitted rep movsq is faster than the vectorized loop.