https://gcc.gnu.org/bugzilla/show_bug.cgi?id=78379
--- Comment #17 from Jerry DeLisle <jvdelisle at gcc dot gnu.org> --- On a hunch, this brings it back. $(patsubst %.c,%.lo,$(notdir $(i_matmul_c))): AM_CFLAGS += -ffast-math -ftree-vectorize -funroll-loops --param max-unroll-times=4 -march=native So -march=native fixes it. not quite as fast as -prefer-avx128, but close enough