https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80697
--- Comment #3 from Michael Meissner <meissner at gcc dot gnu.org> --- Created attachment 41342 --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=41342&action=edit Preprocessed C source for the mult_adj_su3_mat_vec function To reproduce, you need to use the -O3 -ffast-math options.