https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84201
--- Comment #6 from Richard Biener <rguenth at gcc dot gnu.org> --- If Martins bisection to power.fppized.o is correct you can bisect the loop via the vect_loop or vect_slp debug counters (or first try with just -fno-tree-{loop,slp}-vectorize to narrow down to loop vs. BB vectorization).