1996-02-28 - But liars sure can…

Header Data

From: “A. Padgett Peterson P.E. Information Security” <PADGETT@hobbes.orl.mmc.com>
To: cypherpunks@toad.com
Message Hash: f106f99604dbe48fb14114b3d390cff07644ab8096b21b794e9f8dd5da2b338a
Message ID: <960228104510.2022bd44@hobbes.orl.mmc.com>
Reply To: N/A
UTC Datetime: 1996-02-28 18:22:48 UTC
Raw Date: Thu, 29 Feb 1996 02:22:48 +0800

Raw message

From: "A. Padgett Peterson P.E. Information Security" <PADGETT@hobbes.orl.mmc.com>
Date: Thu, 29 Feb 1996 02:22:48 +0800
To: cypherpunks@toad.com
Subject: But liars sure can...
Message-ID: <960228104510.2022bd44@hobbes.orl.mmc.com>
MIME-Version: 1.0
Content-Type: text/plain


I rote:
>>Finally, no cost is allocated to the sustem required to program/evaluate
>>the ponderings of these 100's of ASICs. As anyone who has ever programmed
>>a massively parallel computer (which is what they are talking about in their
>>brute force machine, it is the boundary communications that kill you.

Bill wreplied:
>And the Inmos Transputer, with its 4 communications channels, would do
>just fine as well - you could either build the things into large grids
>or build a large ternary tree, or get fancy and build cubes of 6 chips
>that you make hypercubes out of, or whatever.  A modern version of the same
>chip, built out of FPGAs or ASICs with enough memory and program on chip,
>would blaze just fine.

Oh I agree it *can* be done (thouse who follow RISKS and SCI.CRYPT will note
that I was talking about cascadable single bit boolean processors and DSPs
for brute force engines several years ago. We seem to be miscommunicating
a bit though. I agree that such an engine is easier to design than something 
like a MASPAR or Alliance since each set of kerchunkers can operate 
independantly once started and all that is necessary is to be able to
signal success and pass the found code.

Still, the logistics of initializing each processor set (and ones I have
dealt with have primarily been matricies of single bit devices), providing
communications, power, and control is a non-trivial task: once you have
a general purpose ASIC, you are only half way there, particularly if it is
to be usable for more than one fixed algorithm/key length.

Will further stipulate that a single PC (8088 even) would probably be 
sufficient to load/start/retrieve data from such a beast but design of the
I/O would still need to be accomplished. Not difficult, just necessary.

My point was not that it couldn't be done, but that the costs mentioned
were probably off by a factor of at least two and that for a 40 bit code,
setup for each run would probably take longer than the cracking ( assuming
a court order from Podunk could even get in the queue for a U$300 million
device in an even more expensive facility - students in a University will
*always* be able to do things faster/cheaper than a corporation/government).

Doing something *once* is a lot different from putting it into production.

						Warmly,
							Padgett





Thread