I know, I know - "If P=NP" is a really large, high-impact assumption.
But this is a hypothetical.
I mean, clearly RSA (and similar methods of obfuscation) would likely become totally irrelevant - or would they? Being solvable in polynomic as opposed to exponential time would be a significant blow to their operation, but could they make use of public/private keys so large (if P=NP, then more efficient prime-number-locators seem likely) that cracking them is still something "difficult?"
Those are mostly just points of discussion. In terms of raw question: What methods of cryptography don't ultimately rely on P!=NP?