[109056] in Cypherpunks
Re: DBS Symposium: Going once? Going twice?
daemon@ATHENA.MIT.EDU (Bill Stewart)
Tue Mar 9 18:54:36 1999
Date: Tue, 09 Mar 1999 08:16:10 -0800
To: Robert Hettinga <rah@shipwright.com>
From: Bill Stewart <bill.stewart@pobox.com>
Cc: cypherpunks@cyberpass.net
In-Reply-To: <v04020a09b3098cc9c44e@[139.167.130.246]>
Reply-To: Bill Stewart <bill.stewart@pobox.com>
At 10:16 AM 3/8/99 -0500, Robert Hettinga wrote:
>Even more fun, for the Corporation's first product,
>MicroMint-based microcash, I'm looking at buying the equivalent of
>a financial cryptography supercomputer,
>something I'm calling an underwriting engine.
Out of curiousity, what kind of computation do you need to do lots of?
Is this one of the hash-based algorithms, or are you looking
for DES or modular exponentiation in mass quantities?
There are some modexp accelerator boards from people like Rainbow,
and there have been some people (maybe Eric Hughes?)
who've done moderately large FPGA machines using Altera or Xilinx
which might be better for hashing. [Sigh - too bad the Transputer
is so far in the past :-)]
Deep Crack probably could be commercialized, though for
financial system you'd probably want a somewhat different design.
The underlying DES ASIC core was designed for DESing lots of bits,
rather than for processing lots of key schedules,
and was adapted into a cracking machine using control and data flow
that were designed for distributing lots of deterministically
different keys to crackers, tracking what parts were alive or dead,
and collecting a lot of 0 bits with very few 1 bits mixed in.
Thanks!
Bill
Bill Stewart, bill.stewart@pobox.com
PGP Fingerprint D454 E202 CBC8 40BF 3C85 B884 0ABE 4639
