From: "Peter Trei" <trei@process.com>
Date: Fri, 17 Jan 1997 12:07:38 -0800 (PST)
To: cryptography@c2.net
Subject: DES Key Recovery Project, Progress Report #6.
Message-ID: <199701172006.MAA20293@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


DES Key Recovery Project, Progress Report #6.

Well, RSA now has put up the official rules, and some
test data, for the DES Key Recovery Project. See
http://www.rsa.com/rsalabs/97challenge/ if you haven't
already. The actual challenge will be posted 28 Jan, and
the prize will be $10,000. 

I've had some nibbles of interest from the FPGA community,
but not too much yet. If you know anyone with FPGA or ASIC
hardware handy, tell them about the project, and point them
at Wiener's paper (it can be found at 
http://www.aist-nara.ac.jp/Security/doc/ among locations.

My software is complete, except for one thing: I have not
incorporated Svend Olaf Mikkelsen's new Pentium code
(see http://inet.uni-c.dk/~svolaf/des.htm). This is about
25% faster than mine, and well worth the extra effort. The
conversion turns out to be non-trivial; it uses a different
initial and final perm, the key schedules look different, and
the rounds seem to carry a lot of context from one round to the
next in the registers. All but the final problem is not too
important - the data are all different by simple permutations,
and the conversion is not too hard.

The final problem is more interesting, since I do the first
DES round only part of the time. This means that I'll have to
do more context setup for those times when I'm starting at 
the second round (most of the time), and preserve that extra
context after the times I *do* run the first round.

The other problem is that the 'half-matches' and penultimate
round half-matches will be different for the two systems. This
complicates results checking.

Anyway, come Monday, I want to email to a few people the 
generic 'C' code, along with my assembler version. If you're
interested in doing a serious critique and/or porting of the 
code, and are a US/Canadian citizen living in the US/Canada, 
then email me with a statement to that effect. I'm only sending
out a few copies - this is a slow beta, after all.

If I don't get the Mikkelsen code in there by Monday, it 
should follow soon thereafter.

After the DES Challenge is underway, I may try looking at
the RC5 Challenges.

Who else will be attending Verisign Partner Days or the 
RSA Data Security Conference at the end of the month? I'd
like to meet some of the other participants on this list.

Peter Trei
trei@process.com