https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126371
--- Comment #3 from Filip Kastl <pheeck at gcc dot gnu.org> --- It is also possible to replicate the slowdown with -march=native. I think that for Neoverse N1, it's the same as -march=armv8.2-a+crypto+fp16+rcpc+dotprod+ssbs.
