https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71232
--- Comment #2 from Eric Botcazou <ebotcazou at gcc dot gnu.org> --- > Not sure if the testcases are really having too low loop trip count or the > change is causing a real-life regression (likely). The testcases are quite artificial so the former is not to be excluded, but that only 2 testcases fail would indeed point to the latter.
