[rc5] Next Crack
STROTTIER at novell.com
Thu Aug 21 15:34:42 EDT 1997
I'm not sure why you think that the Bovine effort would be interested in
RC5-64. In the FAQ, they mention Mersenne Primes and Distributed Chess
Engine as candidates for the next project but do not mention RC5-64. Both
the prime numbers and the chess engine sound interesting. RC5-56 is only
interesting because it is solvable within a year. RC5-64 has 256 times as
much keyspace and would likely take much much longer to solve than RC5-56.
I don't know about the others participating in this effort, but I would be
unlikely to spin my CPU on a project that I thought was going to take 20
years to finish.
strottier at novell.com
>>> Fedor Kouranov <ted99 at ibm.net> 08/21/97 01:08PM >>>
While everybody are discussing the projected new & improved clients, I have
some thoughts about coordinating worldwide efforts on RC5-64 (BTW, consider
these thoughts as (C) 1997 distributed.net).
I am totally sure that as soon as we crack RC5-56 many guys on the net will
jump into RC5-64. Probably they will make quite a few efforts (since the
keyspace is _big_ and probability is equal). What do we do to stop efforts'
working on same keys? Coordinate efforts! I think that distributed.net
would shine for this task. What it takes is registering efforts, assigning
keyspace to them and making sure that they don't do double work.
OK, let's coin some new terms. A piece of keyspace that an effort is
searching shall be the effort's sector (space->sector->block->key). The
entity coordinating this shall be, um, keyspace coordinator. The
coordinator will, of course, have no share in the prize and will simply
provide those who register with keyspace that will be unlikely to be
checked by others.
Technical issues: how big a sector should be (how many sectors do we need)?
2^56*256? 2^48*65k? There can be many many efforts - who will be assigned
sectors? Anyone with a PC or only those who can demonstrate sufficient
computing power/distributed support to crack 48/56 bits? How should they
demonstrate this? BTW, I don't want to put too much strain on Duncan ;-)
A "soft-sectored" version: assign only starting points, divide & conquer.
E.g. Bovine starts at 0000..., Cyberian 8000, foobar 4000, barfoo C000,
boofar 2000. Or: Bovine 0000... and up, Cyberian FFFF... and down, foobar -
8000^, barfoo - 7FFFv, boofar - 4000^ etc. When two efforts "meet" they
will be assigned different sectors. This will take some work, but it's
totally flexible. Problem: forcing them into linear distribution which will
weaken security. Solution: inform efforts about keyspace they have in front
and the probability of its being halved.
Ethical issues: should the assigned keyspace be kept in secret? Should we
verify that the efforts are actually doing work or just doing DoS on us?
Waiting for comments.
/** Christ Is Risen ! *** __+__ ****** Fedor "Ted" Kouranov *****/
/* Xristos Voskrese ! ** \| ** ted99 at ibm.net * fedor at bu.edu **/
/** Xristos Anesti ! **** |\ ** http://enz.siobc.ras.ru/~fedor */
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in
To unsubscribe, send email to majordomo at llamas.net with 'unsubscribe rc5' in the body.
More information about the rc5