using braid groups vs. number theory for pki

I'm considering developing an asm prototype of a braid group-based encryption/decryption pki engine. The basic story behind braid group mathematics seems to imply much faster encryption/decryption engines. Before I spend the next month working through this, does anyone know of a braid group-based encryption/decryption engine? Did it work? Any cryptanalysis on this? Also, does anyone have any stats on the performance of popular pki systems? I'm specifically looking for #s regarding number of keys which can be generated by one vendor vs. another vendor's products (on similar hw platforms.) Any clue where I might find this info (other than by paying Giga or Meta $3K for a research report)? Cheers- Phillip