Re: What's Next?

Darrell Kindred (dkindred@cmu.edu)
Wed, 18 Jun 1997 21:22:34 -0400


Colin L. Hildinger writes:
> Actually, the RC5 algorythm is tougher, plus the DES clients checked
> pairs (so we really had a 55 bit keyspace), so for every key it
> checked, it eliminated a second automatically.

This isn't quite true. The complementary-key property of
DES allows you to do half as many key-scheduling operations
but not half as many encryptions, so the complementary key
is not quite "free." In fact, for the bitslice clients, the
complementary-key property doesn't help at all since
the key scheduling cost is negligible.

> For this reason, I expect the RC5 client to not excede
> half the speed of DESCHALL. However, there is a new RC5
> client expected this week for x86 machines.

I'm only vaguely familiar with RC5, so I won't make any
guesses at how fast RC5 key-search could be.

- Darrell

--VAA00472.866685188/gatekeeper.megasoft.com--