Re: Monster machines vs. desktops

C Matthew Curtin (
Wed, 18 Jun 1997 18:13:52 -0400 (EDT)

>>>>> "Greg" == Greg Trotter <> writes:

Greg> But is still would be
Greg> interesting to know: if I were to go buy a machine today,
Greg> specifically to do deschall, how would I get the most bang for
Greg> my buck?

Not a general purpose computer of any type. Despite the fact that we
were able to get pretty amazing speeds through a combination of
Darrell's bitslicing and Rocke's S-box constructions, the speed that
we got would pale by comparison to hardware implementations.

FPGA stuff can be had pretty cheaply; it wouldn't be a big deal for
someone to get ahold of one and wire it for DES. I don't have recent
calculations available, but we could probably search the keyspace in a
few weeks by throwing a few hundred thousand dollars' worth of
hardware at it. Spending even more money, it could be reduced
further. The figure $300 million has been noted a few times in what
it would cost to recover DES keys in a few seconds each.

While it is certainly silly to spend that much money to find the
contents of one message, such a machine would be able to find the
contents of _any_ message encrypted with DES very quickly. Find
enough customers who will pay for this service, and such a large
expenditure suddenly seems more and more feasible...

Matt Curtin  Chief Scientist Megasoft Online    I speak only for myself
Pull AGIS.NET's plug!   Crack DES NOW!