https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61247
--- Comment #7 from Richard Biener <rguenth at gcc dot gnu.org> --- The earlier variants have the known issue that SCEV itself cannot register additional 'assumptions', niter analysis does not need any there, but the fact that 'i' and 'j' don't wrap in the comment#4 testcase does not help to prove that i*N+j does not. We need stronger constraints on N for this.