IP: final note on History of 8008?

[Dave Farber <dave@farber.net>]

------ Forwarded Message
From: gep2@terabites.com
Date: Sun, 12 May 2002 18:59:36 -0500
To: Dave Farber <dave@farber.net>
Subject: Re: FW: IP: WORTH READING History of 8008?

Several recent comments and other messages on this thread that probably need
to 
be added to the historical record we're creating here.

[Re: IP: The 8008 and the AL1]

FWIW, I think that one of the first complete computer systems actually based
on the Intel 8008 was the seldom-mentioned system produced by a NY (Long
Island?) company called Q1 Corporation.  This was a desk-sized system, as I
recall. 

[RE: also  WORTH READING History of 8008?]

I got a couple of notes from Jonathan Schmidt, who was Datapoint's first
Director of Software Development and worked under Vic Poor.  Jonathan went
on to head the Advanced Product Development section of R&D for the company.

I believe I remember hearing that Jonathan was the person who introduced Vic
Poor and Harry Pyle to each other (and they all met through a shared
interest in ham radio).

Anyhow, since these add more interesting information to this history, I
think they need to be added to the thread!  :-)

[quote]

From: "JONATHAN SCHMIDT" <jonathan@bulletinservices.com>
To: <gep2@terabites.com>
Subject: RE: also  WORTH READING History of 8008?
Date: Sat, 11 May 2002 18:02:07 -0500

Gordon, everyone's memory is 30 years old.

Vic and Harry designed the 8008 architecture before joining
Datapoint (Computer Terminal).  I was there the Thanksgiving
of '69 when a lot of it was done.  The 1-bit CPU and recirculating
memory was an expansion of the controllers we had been building
for several years at Frederick Electronics.  Stan Kline actually
made the acoustic delay line that was the memory for those.  Stan,
Vic, Harry (summers), and I were all there together.

My first task was to write a simulator for that instruction set
and parallel I/O bus on an HP 2116.  Harry wrote the assembler
while at Case and I used that.  It worked.

Some more clarifications:

There was no Mike Green until quite a while later.

Vic took Intel into the 8008 architecture and real programmable
computers kicking and screaming.  It was his and Harry's architecture
from a previous life.

TI did make a working 8008 before Intel and Gary Asbell made it work.
It had a very narrow voltage and temperature range, within 50 millivolts,
as I recall, and was never really a product before TI abandoned it.  Gary
may still have an instance of that chip.

and, who cares?

. . . j o n a t h a n

[end quote]

Here's another note I got from Jonathan Schmidt.  While not so much directly
based on the 8008-8080 and microprocessor issue, there's a lot here about
the early development of software at Datapoint, both for the original 2200
and
subsequent.  Most of this is stuff I haven't seen publicly written down
before, so (again) I'd like to see this stuff finally on the record.

Of all the items mentioned below, perhaps AIM is the one neat Datapoint
invention most curiously forgotten by the rest of the industry.

[quote]

From: "JONATHAN SCHMIDT" <jonathan@bulletinservices.com>
To: <gep2@terabites.com>
Subject: RE: also  WORTH READING History of 8008?
Date: Sun, 12 May 2002 00:55:42 -0500

Gordon,
I hired them all and ran the development.  You know that.  I met you at a
conference with a brown bag full of amazing things you made.  You were hired
on 
the spot.

Harry wrote it all:  CTOS, the mag tape OS, and DOS, and, with Klavs
Landberg, the RMS.  Actually, Lee Greenspon did a lot of MagTapeOS under
Harry's 
supervision.  You wrote ARCNET start to finish.  Harry put the "Disk buffer
memory test" in the idle times of the DOS to prove that the disk controller
was 
flaky.  Phil Ray was harrassed by the customer service/mfg rednecks and came
back with the statement that Harry was over taxing the memory...beyond it's
capacity to be correct.  Vic put a stop to that nonsense and the controller
was 
fixed and Harry exonerated.

Mac (bless his departed soul) developed the sync com adaptor before scooting
to TRW in the UK.

The redneck developers all came to work at 11:00 and left for LaFonda to
drink at noon.  Gary Asbell developed most of the hardware from 1971
onwards.  
He was a teenager and didn't know that working was bad.  Since he did a
great 
job, they left him alone to do it all.

Harry, Vic, and I sat in Vic's office and architected Databus.  Vic did the
majority of the work and it was his idea.  There were several "Stath"
routines. 
 Databus had it's own.  I "modified" the string arithmetic to make things
like 
"add one" to be hard coded so that the tests that the 16-bit comptitive
Databus 
bakeoffs would let us come out on top...which we did after that.  Databus
and 
it's ARCNET multi-computer, race-condition-satisfied, deadly-embrace-immune
operation was what put Datapoint really on the business map. Thanks, Gordon,
that was seminal work that still is unmatched....except for Murphy's
multi-computer, multi-user Multiplan that still is unmatched for parallel
group 
participation on a single spreadsheed...that was pretty neat, too, but not
what 
you did.

Gene's 100-times-as-fast FastSort was also a major factor.  So was Murf's
and my AIM.  I did the matrix for random access and Murf showed how to
invert 
it.

Mike's major contribution was writing Pascal and getting us into a high
level language... in the mid-late '70s....then his eyes went bad and he
became a 
country doctor.  What a fellow!

We learned a lot from each other.  The Com-bus of 1972 was certainly a
teaching 
ground for a lot of ARCNET hardware.

We were blessed.

. . . j o n a t h a n

[end quote]


Since we're getting some of this onto the record here... Here's a note from
John Cole, who with Mike Green was the developer of Datapoint's later disk
operating system DOS.H for the 1500 (again, that was the low-cost computer
system from Datapoint which broke with company tradition to use the Zilog
Z80
microprocessor).

His point is well taken.  The 2200 CPU had two sets of registers, ALPHA and
BETA (and similarly named instructions to select the register bank).
Background
processing ran in ALPHA mode and the alternate BETA set was used for
interrupt 
handling (this was quite important given the slow memory speeds of the Type
1's shift register memories!  The [sole!] timer interrupt occurred at a 1kHz
rate.)

[quote]

Date: Sat, 11 May 2002 15:20:23 -0500
To: gep2@terabites.com
From: John Cole <jcole@airmail.net>
Subject: Re: IP: WORTH READING History of 8008?

Hi Gordon,

*VERY* interesting article.  Thanks for copying me on this.  I'd like
to point out a few things, not so much on the history as on the
architectures, though.

Another difference between the 2200 and the 8008 is the "alpha"
and "beta" instructions, which switch to a second set of
registers.  These were used in the task scheduler, as I recall,
and not much of anywhere else.

These instructions, in turn, are almost certainly an artifact of the
Univac 1108 instruction set, which is the only other machine I have
ever seen that has a dual set of registers.  Harry Pyle would have
known this because he went to Case Western Reserve, which had one
at the time.  (I know about it because Illinois Institute of
Technology, where I went, also had an 1108.  Besides, I just like
knowing about hardware architectures.)

John

[end quote]


Gordon Peterson                  http://personal.terabites.com/
1977-2002  Twenty-fifth anniversary year of Local Area Networking!
Support the Anti-SPAM Amendment!  Join at http://www.cauce.org/
12/19/98: Partisan Republicans scornfully ignore the voters they
"represent".
12/09/00: the date the Republican Party took down democracy in America.




------ End of Forwarded Message

For archives see:
http://www.interesting-people.org/archives/interesting-people/