Currently Clang has __builtin_bswap{16,32,64}. It would be great if there was a __builtin_bswapg that works on any integral types that has a multiple of 16 bits as well as a single byte, which would simply be the identity. This would avoid a long if constexpr chain when implementing std::byteswap, which uses the same codegen path in the end anyways. It would also make it trivial to add support for _BitInt as well.