On a more esoteric note: If you are looking to resist quantum algorithms attacks, there are post-quantum algorithms for that[1] (they are computed on normal machines, but the problems behind the crypstosystems are hard to solve even for quantum computers).
SRP also has tunable work-factor knobs, although they aren't as explicit as the ones in bcrypt, scrypt, or PBKDF2. But I'd strongly recommend against re-implementing SRP, since it's treacherous to get right.
Of the implementations listed in the WP article, which, if any, are suitable for general purpose use? And what manner of implementation screwups are most likely|hazardous?
[With the obvious caveat: Advice not for production use, If you have to ask...]
If you're going to use an SRP library, there's an open-source Stanford SRP library (most commercial systems derive from it) which deals with the obvious attacks.
On a more esoteric note: If you are looking to resist quantum algorithms attacks, there are post-quantum algorithms for that[1] (they are computed on normal machines, but the problems behind the crypstosystems are hard to solve even for quantum computers).
[1] http://crypto.stackexchange.com/questions/494/what-is-the-po...