https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119244
--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> --- For __int128 it is all 32-bit targets at least (perhaps with exception of gcn, don't remember). We simply don't have quad-word math support in libgcc/code generation (except for BITINT_TYPE).
