https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84201
--- Comment #16 from Richard Biener <rguenth at gcc dot gnu.org> --- Created attachment 52582 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52582&action=edit simple --param patch This adds the ability to turn off floating point induction vectorization with --param vect-induction-float=0 which should be usable ontop of -Ofast. Can somebody with unaltered sources verify if that helps?
