Re: Optimized key rates?

C Matthew Curtin (cmcurtin@research.megasoft.com)
Sat, 3 May 1997 23:34:54 -0400 (EDT)


>>>>> "Corbett" == Corbett J Klempay <cklempay@hops.cs.jhu.edu> writes:

Corbett> Cool! Can someone provide some quick info on the Biham
Corbett> bit-slicing approach to me, or at least tell me where to go?

I can't get to his site right now to find you a specific URL, but I
can get you the index of Biham's publications page. Finding info on
his superfast DES implementation for 64-bit processors shouldn't be
tough from here, once the server is up...

http://www.cs.technion.ac.il/~biham/publications.html

Corbett> Will Biham's method only work for 64-bit processors (or will
Corbett> it work on 32-bit, but with greatly reduced efficiency?)?
Corbett> Thanks.

This will only work on 64-bit processors, and be at all useful. The
crux of the matter is that DES is a block cipher that works in 64-bit
blocks[*]. Hence, each block can be processed by a 64-bit processor as
if it is 64 one-bit processors. I suppose one could break the block
in half, and double the number of cycles needed to process it.
However, I don't know how this compares (speedwise) to other fast
approaches for 32-bit processors. It might be a performance hit.

* DES, though commonly called a 56-bit cipher, is really a 64-bit
cipher. However, the key 64 bit key is typically broken up into 56
secret bits and 8 bits for parity. Permutations of DES are
possible, then, that keep all 64 bits private, and a stronger DES is
theoretically the result. (I'm not an expert on DES internals,
though, so I don't know what the tradeoffs are here, and what
problems might be introduced by doing this sort of thing.)

-- 
Matt Curtin  Chief Scientist Megasoft Online  cmcurtin@research.megasoft.com
http://www.research.megasoft.com/people/cmcurtin/    I speak only for myself
Death to small keys.  Crack DES NOW!   http://www.frii.com/~rcv/deschall.htm