https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126935
--- Comment #4 from Drea Pinski <pinskia at gcc dot gnu.org> --- (In reply to ktkachov from comment #3) > (In reply to Drea Pinski from comment #2) > > Actually I think the better check is if the processor has SVE and SVE2. > > Even though SVE2 should be enough here; it seems like some VMs advertise > > SVE2 but not SVE support (which in itself odd but that ever). > > Oof, SVE2 but not SVE is a nonsense configuration. > Anyway, for the libcpp fast-path I've added I think we can make the > configure check also execute the SVE2 instructions required, not just test > their HWCAPs. That would sidestep the dodgy VM advertising But then that would break cross.
