From: bill payne <billp@nmol.com>
Date: Tue, 31 Mar 1998 08:31:07 -0800 (PST)
To: DJALLEN@sandia.gov
Subject: T 1563 and pals
Message-ID: <3521072D.7BE4@nmol.com>
MIME-Version: 1.0
Content-Type: text/plain


Tuesday 3/31/98 7:39 AM

John Gilmore
John Young
J Orlin Grabbe

Reading the stuff on the pals at

http://www.jya.com/pal-nukes.htm

http://www.research.att.com/~smb/nsam-160/pal.html

jogged my memory.

After we completed the Forth implementation of the 
Missile Secure Cryptographic Unit at least Hal Pruett,
Duane DeWerff, myself, and perhaps several others
were invited to NSA at Fort Meade in about 1985-6.

NSA employee Bill Legato assailed me for using Forth
for the implementation of the MSCU.

Legato hated Forth and relayed his hate to Sandia management.

This is one reason I was replaced as project leader by Tom Evans
who was later replaced by Eric Disch.

Possibly surprising, however, is that NSA sent Forth expert Don Simard
to
visit Sandia for several years.

Kent Parsons commented to me that he thought Simard was an NSA spy
sent to see what we were up to at Sandia.

Simard, along with Bill Goldrick, were the two reviewers of my 
SAND report now seen at http://jya.com/whpfiles.htm

EE Times, March 30, 1998 page 8

  A Java chip being to ship in quantity
  SAN FRANCISCO - Patriot Scientific Corp. surged to the head
  of the silicon pack at the JavaOne conference last week, showing 
  the PSC1000 RISC processor, which is shipping in quantity.
    "We'll be selling the chip for under $10 in volume and aiming at
  deeply embedded, low-power applications such as handheld devices
  and factory-floor control systems, " said Phil Morettini, vice
  president of Patriot (San Diego).
    Competing offering from LG Semicon and NEC Corp., each based on 
  Sun Microsystems Inc.'s picoJava core, are not yet available, although
  select customers are said to have samples.
    Separately, Rockwell Collins Inc. (Cedar Rapids, Iowa) said that it
has
  decided not to sell its Java-specific CPU on the merchant market.  The
  company unveiled the innovative processor last fall.

Here is their web site.

http://www.ptsc.com/

Charles Moore, inventor of Forth, is author of the Shboom Java native
code compiler for the PCS100.

Rockwell built a Forth-like chip many years ago.

I met Randy Dumse, who then worked for Rockwell, at a trade show.

I got a copy of the Rockwell R65F11 chip manual from Dumse.

Dumse quit Rockwell to start NewMicros seen at 

http://www.newmicros.com/

Click on history to see

http://www.newmicros.com/history.html

Of course, the Harris RTX 2000 chip used by NASA is a descendant of
Charles Moore's Novix Forth chip. http://groucho.gsfc.nasa.gov/forth/

Ballard, I was told by Jeff Allsup of Wood Hole, used polyForth to
discover the wrecks of the Titanic and Bismarck.

A central debate is whether a high-level language should be used for
a weapons programmer.

The implementer must be able to account for each bit.  In Forth I can.

Possibility exists that an disgruntled implementer might 'spike' a 
weapon so that it went off at an unintended time.  

Sandia supervisor Don Schroeder was an OUTSPOKEN critic of the T 1563
nuclear bomb controller.

SCHROEDER,DONALD H.       (505)845-8409  DHSCHRO   (505)844-9478  0519
DHSCHRO@sandia.gov

The T 1563 did not work very well.  One reason is that it was written in
about 112,000 lines of assembler.

NSA was reluctant to approve use of an operating system or even a
high-level 
language for crypto implementation for the reasons that security could
not be
guaranteed. 

Sandia did not have the resources required to write its own secure
operating system
and high-level language.

This is why we used the Forth operating system and high-level language.

Schroeder division employee Mike Sharp

SHARP,MICHAEL W.          (505)845-8444  MWSHARP   (505)284-3850  1138
MWSHARP@sandia.gov

wrote a T 1563 implementation in, I think C or Pascal.

Henry Newgass modified the transient part of command.com 
so that on power-up a PC directly entered Sharp's program.

Jerry Allen, who also saw the merits of using an operating system
and high-level language, funded Neugass.  

ALLEN,DOUGLAS J.          (505)845-9624  DJALLEN   (505)844-7593  1201  
DJALLEN@sandia.gov

I wrote the purchase order.

I learned several months ago that Sandia built a PC version of 
the T 1563 called the T 1663, I was told, which was supplied to 
the Air Force.

The threaded code - compiled native code battle continues.

I am sort-of happy to be back [somewhat] doing MASM code -
along with a bit of Forth and Forth assembler on the 80c32.

I see that J Orlin Grabbe is linking

  Permissive Action Links

at http://www.aci.net/kalliste/

I read

    Both strong links must be closed electrically -- one by specific
operator-coded input and one by      environmental input corresponding
to an appropriate flight trajectory -- in order for the weapon to be
armed. 

George Dullek and Jon Bryan worked together on accellerometers for
flight trajectory sensing.  

Dullek and Bryan are Forth programmers.

DULLECK JR.,GEORGE R.     (505)844-2628  GRDULLE   (505)844-8480  1073
GRDULLE@sandia.gov

BRYAN,JON R.              (505)844-2015  JRBRYAN   (505)844-6161  1003 
JRBRYAN@sandia.gov

Sandia had a division devoted to crystal oscillators.

The oscillators for the micro occasionally fail to start - a bad scene
if this happens when trying to arm
a nuke.

Dick Adams, also in department 2300, was in charge of oscillators. 
Adams never seemed to many of us
too interested in his job.

Adams took an early voluntary separation package from Sandia with a
large educational retraining
incentive.  And went to cooking school in San Francisco, Bob Wayland
told me.  The real world again.

Thanks to good research, some whistleblowing, and Internet details of
fuzing the US nuclear arsenal are being revealed.
  
Let's ALL hope for SETTLEMENT of this unfortunate mess before it gets
WORSE.  

I want my money and I am out of this mess.

Later
bill