1995-12-18 - Re: What ever happened to… Cray Comp/NSA co-development

Header Data

From: anon-remailer@utopia.hacktic.nl (Anonymous)
To: cypherpunks@toad.com
Message Hash: c8c207d01f1e450ee00e5cefd476aed31d88f16f05099596e9ac9938eaeedbc9
Message ID: <199512182241.RAA01276@bb.hks.net>
Reply To: N/A
UTC Datetime: 1995-12-18 23:37:15 UTC
Raw Date: Tue, 19 Dec 1995 07:37:15 +0800

Raw message

From: anon-remailer@utopia.hacktic.nl (Anonymous)
Date: Tue, 19 Dec 1995 07:37:15 +0800
To: cypherpunks@toad.com
Subject: Re: What ever happened to... Cray Comp/NSA co-development
Message-ID: <199512182241.RAA01276@bb.hks.net>
MIME-Version: 1.0
Content-Type: text/plain


-----BEGIN PGP SIGNED MESSAGE-----

At 09:15 PM 12/17/95 -0500, Thaddeus J. Beier wrote:
> ----------
> ... about the fate and capabilities of the CCC PIM
> (processor-in-memory) machine.  A friend of mine was
> working on it, and it would have been a screaming machine,
> no doubt about it.  He said that the Cray mostly acted like
> a really fast network for the processor chips.

I had thought that it may have been similar to an Active Memory
design, but had no real clues to go by. Wasn't sure about the
chip array path width either, but thought that 1 bit was probably
the way to go given the array size.

> ... the PIM chips were made by a dedicated NSA company,
> Supercomputer Research Center, in Bowie MD.

Hmm... is that the name?

> But, it was nowhere near finished when the company finally went
> down, and the team was completely disbanded.  My friend was talking
> about going to the auction when the parts of the various machines
> were going to be sold, I don't know if he did so.  He suspected that
> the various pieces would end up going back east to the Fort
> Meade area.  Still, it is such an odd machine that you would
> probably have to transfer the staff to finish it, and that didn't
> happen.

Sorry Thad... but NOTHING 'just disappears' at the NSA...

> In any case, while it was fast (1/2 million 1-bit processors,
> perhaps as low as 1 nanosecond (1 GHz) cycle time), it was not fast
> enough to brute force reasonably strong ciphers.  It's really no joke
> that it would take a computer with picosecond clocks the size of the > earth more than the age of the universe to brute force IDEA, for
> instance.

Hahahahahahahah!!!!!!!!!! A cryptographer's most POWERFUL weapon, is
A False Sense of Security... If you've read Kocher's abstract, you
would NO DOUBT realize that there are ALWAYS options to be explored
and exploited. I'm sorry, don't take it personal, but I think that this
'til the end o' time' argument leaks like a "sieve"... it's the classic
linear thinking thing.

> It would have made a great DES cracker, though; my
> back-of-the-envelope calculation has it cracking one key every
> .75 days on the average.

This sounds pretty pessimistic too...

> thad

> -- Thaddeus Beier                   email:  thad@hammerhead.com
>    Technology Development             vox:  408) 286-3376
>    Hammerhead Productions             fax:  408) 292-2244
> 
> ----------

Well... maybe the timing wasn't right for the system's completion.
After all, designing a computer with <1 nanosecond cycle time is not
child's play. And in quantity it is even more difficult. A few more
years of development with Transphaser logic and Holographic storage
could work some wonders in computational capailities.

Anitro





- ---
[This message has been signed by an auto-signing service.  A valid signature
means only that it has been received at the address corresponding to the
signature and forwarded.]

-----BEGIN PGP SIGNATURE-----
Version: 2.6.2
Comment: Gratis auto-signing service

iQBFAwUBMNXt+SoZzwIn1bdtAQEQAQF/YKCYcSiH4BRswP7+4Sv1VIYynpH738vF
LFa/31pQBJiEkmkpiykmOcL3YySghGsf
=3nUQ
-----END PGP SIGNATURE-----





Thread