https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122058
--- Comment #23 from Sam James <sjames at gcc dot gnu.org> --- Diffing the dumps on trunk with -fno-devirtualize on icu::number::impl::NumberRangeFormatterImpl::resolveModifierPlurals is very noisy and not sure it's very helpful. vrp* differs but that's obvious because we don't have an inlined copy to analyse then. fre1 has the same issue. I'll look at dumps between the 10/11 commit instead.
