[RC5] Golomb Rulers (was "sexy" projects)

Thomas Womack thomas.womack at merton.oxford.ac.uk
Wed Mar 4 01:28:46 EST 1998


>'Thomas Womack' said previously:
>| >     It has been calculated that a direct attack on dedicated hardware
>| >     could for ~US$10,000 break DES in a few minutes.
>|
>| Not anywhere I've seen; it's almost impossible to get $10,000 worth of
>| custom-made chips (minimum orders being a lot bigger), and, for an attack
to
>| take 2^9 seconds ('a few minutes'), you need 2^17 250MHz crackers. Not a
>| $10,000 project by any means.

>Take a look at "http://www.isi.edu/mosis/" and go download Magic and
>Spice 3 at your local FTP site.  Design rules and sample pad frame are
>available for download too, if I recall correctly.  Some Universities
>use this service for their VLSI courses (mine did).

I've read about those; I haven't had the chance to do VLSI design myself,
but I checked the pricing fairly carefully before posting. Yes, it's
possible to do the design for free - but what gets you, and what I can see
no way to avoid, are fabrication costs.

The price of chips through Mosis is approximately $120 each, for the
smallest possible chips done in 0.35u. Even assuming you could fit 16 Weiner
machines on a single chip, which you can't, you'd need 2^13 of them, for a
total cost of about $1 million (and this is assuming you could run them at
250MHz, which you can't).

However you go about it, 2^56 is a big number. The DES attack took
distributed.net, who have five times the processing power of any
commercially available supercomputer and are using idle time on perhaps $50
million worth of hardware, six weeks. Doing it three orders of magnitude
faster for a factor of fifty less cost is impressive - but you need another
two orders of magnitude cost reduction to make hardware approaches at all
cost-effective.

Tom
--
To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list