[HARDWARE] PAM, CeepDrack (filter this :), FPGAs

stoney at sequent.com stoney at sequent.com
Wed Sep 15 13:25:57 EDT 1999

Yes, a former Digital Research Lab in France was doing research
on PAM boards in order to accelerate applications that are not
well supported by general purpose microprocessors.  One application
that they implemented was real time decryption of a DES stream.

The PAM board contains fpgas to implement the bus interface, fpgas
to implement the algorithms that a general purpose CPU does not 
implement well (chinese remainder, etc), and SRAM.


  Schneier, Bruce,
  "Applied Cryptography : Protocols, Algorithms, and Source Code in C"
  John Wiley and Sons 1995

Deep Crack used a low tech ASIC to implement 24 search engines for
a brute force attack to DES.  They put many of these ASICS on a board
and many boards in a modified Sun chassis.  They used a i486 sys to
poll the boards for search engines w/ intresting matches to check w/


They partitioned the problem between software and hardware to increase
their capabilities.  Development of an ASIC is beyond the resources 
of most individuals, but an FPGA solution would be feasible.  You 
wouldn't be able to easily beat DeepCrack, but you would be able to
solve your problem faster and learn a great amount.

Another problem you would have is having multiple layer PCB boards 
fabricated to handle the signal integrity problems you would have w/
bus speeds beyond 10-15MHz.  You might be able to clock the device
faster w/ less layers, but not the busses between them.

Graphics processors are a great solution to graphics problems.  You
don't saturate your processor bus w/ traffic and you allow your proc
to work on more important tasks.  The KEY is not saturating your bus.

Producing graphics boards is a tough business.  The volume is high
enabling cost savings from econimies of scale, but the margin is 
much lower.

Develop a PCI board w/ FPGAs and SDRAM after you have played w/ an 
FPGA development board.


On Wed, 15 Sep 1999, Jameel Akari wrote:

> On Tue, 14 Sep 1999, Zach Edwardson wrote:
> > Would it be possible to develop a Pc card for computers that uses a
> > Inxepensive chip, with say 4-8 megs of ram, to crack Rc5 (or just to be
> 	Certainly it is possible.  You could lay out the RC5 algorithm in
> hardware, all pipelined and superscalar, and it'd haul ass, but there's
> nowhere near enough market to get the price even to $100, let alone $50.
> 	I've thought of doing this with one of the available FPGA boards
> outthere (we have DEC Pamettes and Wildfires, if anyone recognizes those)
> which certainly have enough gate to do it.  But that would require many
> hours of design work to do it right, plus software development (make the
> client work with the cards).  Plus a Pamette runs about $5K a piece...
> ------------
> Jameel Akari
> Insert witty comment here
> ------------
> <http://fribble.cie.rpi.edu/~jakari>
> ICQ: 27182003
> --
> To unsubscribe, send 'unsubscribe hardware' to majordomo at lists.distributed.net

To unsubscribe, send 'unsubscribe hardware' to majordomo at lists.distributed.net

More information about the Hardware mailing list