https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81616
--- Comment #34 from Jan Hubicka <hubicka at ucw dot cz> --- > So gcc loses on mt19937ar.c without -mno-avx2 > But gcc wins big on matrix.c, especially with -mprefer-vector-width=none > -mno-fma It is because llvm does not use vgather at all unless avx512 is present. I will look into the vgather cost model tomorrow. Honza