RE: what does this all mean?

Ethan M. O'Connor (zudark@mit.edu)
Sat, 17 May 1997 17:17:17 -0400


>>it has been argued that design and fab costs notwithstanding, a
>>chip could
>>be built with todays tech to crack a DES key per clock cycle
>>and it could
>>be clocked around 2GHz. this potential chip has been theorized
>>to come out
>>to about $50 retail, if it were built.
>
> Err, while it is true that hardware solutions can be very fast,
>they aren't that fast. I assume that you meant a chip that
>could CHECK a DES key per clock cycle. In fact, DES was
>designed for hardware, and should fit a pipelined system very
>easily.
> I know of no logic technology that can reliably be clocked at
>2Ghz.
> The fabrication cost of a small processor is usually about $15
>depending on the technology (Heck, Pentiums have a die cost of
>around $50) if you don't count the initial masking and design.
>
> Feel free to correct me however...I'm interested in seeing
>literature about a 2Ghz clocked chip...
>
>- ---

Even assuming only a 1Ghz clock cycle (which is reasonable...
this chip would presumably be quite simple...), checking
1E9 keys per second would search the entire key space in
72E15/1E9 = 833days; you could build an entire checker for
< $100 so for $10 million you could set up 100,000 of them
searching different parts of the keyspace and be done in 12 minutes.

That doesn't seem like too much money. If you count design and
tooling costs, it will obviously be higher, but certainly still
well within the reach of any mid-sized corporation or government agency

----------------------------------------------
Ethan O'Connor |Does cracking DES seem like too
zudark@mit.edu |easy a challenge? Join the
|Great Internet Mersenne Prime Search
|http://www.mersenne.org/