Does that mean that if you're running as 32bit and want to store a number larger than that (a double or int64_t or so) or a POD that all of these have to be allocated seperately?