I can't find it right now, but I recall some paper by Ed Fredkin that claimed that measuring fundamental physical constants indicated a 256-bit register could work.

I think the paper was in an IBM in-house technical journal.