Return to On-Line Documents
NOVEMBER 18. 1981 Dick Clayton There are a couple of people I'd like to mention that are here tonight also and it is with great trepidation that I mention some names because I'm bound to miss a few others. Wes will talk about the significance of the people along the way. We have Charlie Molnar who is the engineering designers side kick of Wes and of most of this biomedical computing and a lot of it all the way through. It was giving us a lot to worry about in engineering design along the way and Charlie is really superb. Mary Allen Wilkes Clark, the programmer that lead to much of the base operating software for the LINC and its various versions. Mort Ruderman, he was the salesman that sold Wes the modules. Some other names out of the past there are Norm Kinch who made it all happen and Lion Malpass. Having blown all the rest of the other names there, let me introduce Wes to give the speech. He is clearly one of my very early educators on computer architecture and a very tough task. master around computer design and he is also the youngest of the computer pioneers to come give our lectures. Wes Clark Thank you, Dick. It is a great pleasure as you can tell by the smile on my face to be here tonight and talk with you. I have to say that I never thought of the TX-2 as exactly a simple straight-forward designs but I'll take it in the spirit as it is offered. I have to admit that I put Dick up to mentioning the Eckert-Mauchly Award because at that presentation I made the comment that up until that point after having maintained a couple of decades of verve low profile professional activity my only prior distinction had been the only person that had been fired by MIT three times for insubordination. It wasn't much of a ceremony actually. Not very festive but there is a slide that Doug Clark. took. of the occasion. I did, however, in looking through records to find material to show you tonight to keep your interest; came across this lovely thing which I told Owen about earlier, this is a brochure published in 1966 by the Honeywell Core. The Honeywell Core. commissioned a British cartoonist names Robert Emmet who was published extensively in Punch magazine over the previous decade or twos maybe still at work I don't know. They commissioned him to design a rather festive computers which he did and you can't see its but you're certainly welcome to come and see it afterwards and I do think. the Museum should acquire it. It is called the Honeywell Emmet Forget-Me-Not Computer. It is actually a charming thing. Let me .lust read about it very briefly. The Forget-Me-Not Computer comes in three parts, like Henry VI. Built in strict accordance with Livingston's laws 'Memory may hold the door but elephants never forget", it is a pleasing outward appearance being delicately constructed of bamboo so that it maw be placed in any executive suite without offense. It contains a dazzling array of features. Some well--tried, some absolutely new, and some on the fringe of the future. The machine is designed in the shape of art airy elephant. A very open work. thing, full of gadgets, reminiscent of Rube Goldberg. Let me summarize some of these things. Mass memory - it is characterized by heavy watery he thought that would be sufficiently massive. Also it had a drain, called the brain drain, which facilitated the flushing of any wrong numbers and unworthy thoughts. The speed was measured in Billy cycles that being the length of time it took a little man on a bicycle called Billy to get from one end of the frame to the other. Floating Point unit with electro magnets going to and fro on the water. Memory lane considered by most to be the moving fitment in the entire computer. Memory lane offers specially treasured pieces personal memorabilia including the ancestral home where Honeywell started in 1885r mothers a sweet old thermiotic valve, first love where positive meets negative, etc. He goes on to saw that all the preceding features incorporate the latest solid state design. In the Forget-Me-Not computer solid state takes the form of half a brick from a Scottish stately home. Nothing could be more solid or state than that.
The LINC story the LINC apparatus, the LINC history predates the Honeywell
Emmet Forget-Me-Not
computer. You'll see that it is not made of bamboo, it maw look equally fragile,
and in fact parts of it are
made of wood. The console boxes are plywood. Very
effective stuff. Well for many of you that is gains to be new and what I talk
about tonight and show you will
be new and for many others it will be very much a trip down memory lane and
I hope that you will appreciate
it.
The story actually starts in the mid-50s when we were at work on the TX-0 and
TX-2. The croup later
under Bill Papian was one in which Ken Olsen and I and many others whose names
you will know. Dick
Bests Tom Stockebrand, Ben Gurley, Ed Hartwell, and some others worked in the
development of computer
technology at Lincoln Laboratory. We had through one of our staff members,
Belmont Farleigh who sot in
touch with Walter Rosenblith's laboratory for communications of biophysics
at MIT. Out of which came two
other very important members of the croup; Charles Molnar being the principal.
In the course of interaction
with the croup we became rather well versed in some of the difficulties in
using the computer in bio-medical
research. Well not bio-medical specifically neurophysiological particular
kind. We came into contact with the
principle funding agency of the government during that period which was the
National Institute of Health.
Primarily through contact with Bill Papian, the croup leader with one
Dr. Bruce Waxman of the Division of
Research and Facilities and Resources of NIH.
I didn't think too much of that idea at the time
because that was a toy of mine. I was busy using it as was Bill
Mont, but Bill pointed out rightly that we would
be switching our interests more and more to the TX-2 which
was then in a building in the late 50s. So in
1958 it did indeed transfer down to MIT. I had of course expected
it to do to Walter Rosenblith's laboratory.
He had been the principle user of that instrument at the time. His
group of people had been among the principle users at Lincoln Laboratory as far
as MIT was concerned. So
it was suite a surprise to me that Walter declined, backed away at a very
high speed as a matter of fact. It
was too much for him to handle. He thought it would be too disruptive to
the normal workings and traditions
of that laboratory which was a small and very talented group of people
with many visitors from all over the
worlds and a laboratory of
many important methodologies were developed over the years. As we
know Walter became and went on to
become the Provost of MIT. Well that was his prize.
It certainly had me thinking what was wrong with the TX-0. Although in
the form that it was sent to MIT it
was enriched considerably in order code, but reduced in memory size.
In May of 1961, couple wears later
Walter led a group of us to California to UCLA where we helped to
dedicate the Brain Research Institute at
the Medical School there. In my contribution to that I talked about
digital techniques and their data
processing. I had by the way sort of as a substitute for the TX-0
some kind on instrument that I thought the
Communications Bio-Physics Laboratory at MIT could use well had
put together a logically small gadget
called the 'ARC'. The Average Response Computer to which Dick Clayton
referred to. Small it wasn't. It was
the size of an ark. It was the size of a refrigerator on its side,
and on huge casters. It is the sort of thing that
You would put into a box of this size or perhaps a scope probe
these days. That instrument was used for
many years in the quantification of neuro-electric signals arid
much useful research was done using that tool.
The TX-0 however, has about 8000 registers of digital storage
and is organized as a general purpose device.
It is quite simple to program the TX-0 to act very much like the
ARC,
in fact this has been done requiring an
investment of effort in measurement of only hours. Quite a
familiar story to us, but to the brain researchers
over the periods certainly not. It can more over generate
quite various displays of data and results of analysis
and can be programmed to carry out exceedingly lengthy and
complex operations if desired. This flexibility
and behavior is characteristic of the story of the stored
program computer and obviously a treat value. So at
the end of this I said, that the stored program computers
like the TX-0 are beginning to appear in
commerical form. I had in mind I guess the PDP-1,
There was reason to hope that these machines were perhaps other
general purpose machines with the
capability somewhere between the
ARC
and the TX--0 will find their way into a laboratory. The most
significant benefit promised by such development is the greatly
expanded and freedom of choice of the
various methods of analysis that would be made possible by the
use of these tools of research. That wasn't
especially present of me, but I had already been thinking about
the laboratory instrument computer. Then in
deeds not by that name as it happens. Then as I got back to Boston, I
think. on Mother's Day of 1961. and I proceeded to work on the design
of the thing whatever it was
at the time. I took. an informal leave of absence from Lincoln Laboratory.
I don't know how those of you in the digital design business to about
your thing, but for me it is a matter of
finding some kind of gimmick to start with. It doesn't much matter
what it is as long as it sets me started. It
survives weeks, months, depends on the size and complexity of the
design and also on the soundness of the
gimmick. My gimmick here was what you see above a set of registers
pretty much identical in form, but
identified by function. At the top you see "R" I don't know what
that meant. 'F' was some kind of function
box in which all the processing would be carried down. 'PQ' and down
to 'A' at the bottom that I was
distinguishing as an accumulator as though each of those would
obviously be able to accumulate was at the
end of the chain. I imagine there were eight altogether. Memory
is up at the top that was interchanged to 'R'.
Magtape which I clearly realized that we needed in some form and
in fact thought of a variant of the TX-0
tape system. The busses themselves would have been 3-bit buses.
I was familiar with the DEC system
modules at the time and there were a number of quad units that made
a 12-bit word, that I was convinced
about the right size. The CDC 160 was an existing proof that
at least one other designer thought that it was
practical word length. So three of those quads made a resister
and that was a nice working group. So it was
a nice idea, transferring around the microprogramming actually
the modern idea for 1961.
This is the first picture of the LINC magnetic tape unit. These
are scraps from my notebook by the way, As
I said I had left Lincoln Laboratory rather informally and had
set up at home on the top of a nice pinewooded hill where this
stuff is all done in the spring and summer. Ben Gurlew as a matter
of fact used to
come over at the time and we would talk about the design decisions
that had to be made on the PDP-1 which
he was working on. It seems that I had two in mind, my notebook
which is very scratchy at this time and not
very well shown, but there is the tape in the corner. As a matter
of fact here is a picture, just a couple weeks
later in the end of June, of the second known picture of LINC
tape unit. I think I was beginning to get
some appreciation of the horror of the control that would be
required to carry out the very simple functions
that I had in mind.
For example the control of the tape unit is built into the machine and
under the control of single construction. ???instruction???
It was not a very compelling architecture. but (the entire concept of putting this
in one box that an experimenter could take away to his laboratory
and work with in a personal way was in fact the essence of the thing.)
This is a staff of consoles being operated by William Simon who came to our
laboratory as a trained
physicist, a very brilliant guy, in fact a friend of mine once said he
was the only person he knew that could
give 17 valid reasons why Hudson's Lake should be drained. Quite possibly
one of the brightest minds in biomedical engineering working today. He
is presently at the University of Rochester. A number of other
people, Severil Weinstein<?>, Tom Stockebrand, Merv Allen Wilkes
of course we had help from the
technical staff and Norm pinch in fact he took track of an incredible
number of things that you'll hear about
later. Finally we got together a demonstration prototype and tried it.
Let me show you first of all the kinds of
pictures we were making at that time. This was actually done a
couple of months later as you can see.
Charlie and I took this almost completed demonstration prototype to
Washington in May of 1962 and
demonstrated it at a special conference of the National Academy of Sciences.
By the was Bill Simon set up
an elaborate set of television sets around the auditorium about this size
and it was a very effective
demonstration. I hadn't known until I stepped up to the podium if anything
was going to work. Until 7 minutes
before Charlie was crawling around practically in his underwear either
pre--loading programs or whatever,
had to rush up to the hotel room to change into his uniform. He was a
member of the military and felt that he
had to make a formal appearance. As I set to the podium Charlie put in
front of me a small piece of paper
that said 'The following have my confidence and he named three or four
demonstration programs that had
worked and was absolutely trying
every one that worked.
Following the demonstrations of course everything fell apart.
Charlie recalls one person came up afterwards
very intently studying the machines insulation and we
thought that was rather curious but it turned out he
was from the Smithsonian. I must tell you from there we
took it to the laboratory of Dr. Robert Livingston
who was then the Chief of the Neuroscience branch at NIH
and I can show you the first serious input
device for the LINC. That's Jasper the wonderful laboratory
animal with implanted electrons that was
hooked up to the LINC and from whose auditory
implanted electro signals we extracted responses to
auditory stimuli<?> from the <?> microphonics.
Charlie could fill you in with more details with that later, if
you had any questions.
In any case an open house was held and people from all over NIH
and various labs including the computation
centers came to look at what was going on. We remember (one
fellow looked at the LINC inside and out,
and at this wire going over to the other side and said 'This
thing possible can't work, there is no way to get
the data in'. He couldn't find any punched cards. We took the
things back to Lincoln Laboratory exhausted,
but triumphant and wanted to do more.) Obviously we were just
on the beginning edge of this and we hadn't
even finished the machine vet. The machine was subsequently
finished and put into use by Dr. Tom Sandler
another member of the computing communications bio-physics laboratory
under Walter Rosenblith who had come to Lincoln and was most
anxious as were we to establish a so called
'wet' facility. That is an animal source facility for study of
neuraural event of behavioral processes. In
which we would have used the TX-2 to be proposed building a
large wet laboratory. People in those days
distinguished wet from dry physiologists. The dry ones were
the theoreticians Lincoln Laboratory did not like
the idea. Lincoln Labs told us we couldn't continue that work in
Lincoln and that if we couldn't find a way to
embed our work in some of the ongoing work of some of the defense
programs we would be better of
finding other auspices. So in fact we went on the road for quite
a while trying to find other auspices. A
number of very dedicated people that worked very hard on the LINC
program were elected to go along
and the only question, was where. It was finally in the Fall of
that year (1962) that Walter Rosenblith, that
you see here on the right talking to the present head of the
communications bio-physics laboratory, Dr. Louis
Braida, suggested that in fact there was a bigger thing that
could be put together, something that would
encompass a great deal of computer research and a great deal
of bio-medical research under one roof. He
proposed that it be done with a very large sum of money that
NIH sponsor that and MIT would act as a host
institutions but that any institution within hundreds of miles
of MIT and Harvard with the exception of Yale of
course would be participants.
In parallel with this development of the LINC it was going on
a further establishment of what was then
called the center for computer research in the bio-medical sciences
that Walter Rosenblith was organizing
with Bill Papian. In fact that was funded in a dazzling display of
speed and efficiency by NIB. The
funded almost $30M on a seven year grant. It was the biggest
thing that NIH had ever done and it made the front page of the
Times.
The kit you can see up there has wired frame with a power
supply on the right, a stack of system modules.
By the was not all of those are system modules about 10%
of the circuits were designed by Charlie. They
were better than DEC's at the time by considerable degree,
but also we were concerned about giving DEC
everything there was to give or to have about this project.
We wanted to keep this a more open affair. So
DEC only got 90% of the electronics. You can see the
tape unit those pairs of eyes, under it the scope, the
console on the right, a stack of documents, and a terminal
box sort of plug-ins in which arbitrary apparatus
could be put to couple the machine to laboratory environment.
You can't see because they are black to the
table at the lefty but there is a very big heavy coil of
cables that connect the electronics frame to the finished
product. You will see that when you see the LINC in the
other building later. But that is what we had in
mind. They would come and learn about that thing. Indeed
Tom Sandle had agreed to act as the chairman
of the LINC evaluation committee which was formed, a formally
constituted six member group of very
distinguished scientists who would review
about 72 proposals that were made by people in response to
an invitation published in Science and most
of the university departments around the country. Then
then would pick from them about 12 who would
come to Cambridge in two waves of 6 each for one month
of intensive training in the Summer of 1963.
We taught them and here you see Severil<?:>
with a bunch of the participants. Mary Allen explaining some
points of the programming and logic. The caption
on the posted picture which was taken by Bill Simon was
'and now children the Momma bear said to the Papa
bear...'. Irving Toma, then a graduate student
in physics
came to us, I guess he bounced off the communications
physics laboratory some how and got involved. We
had to teach these people everything as you can see
down to the real fundamentals and the real nitty gritty.
But finally the frames did arrive in good shape and
you see here Dr. Keith Killman and Jim Hance from
Stanford University from the department of pharmacology
looking at the frame, and Jay Walter Woodbury
from the University of Washington in Seattle. (Ken
Olsen used to say in the days that we built them 8 and 10
feet high that it wasn't a computer if you could
see over the top of it.) Jay Walter was very tall and perhaps
it is more than symbolic that he had some of
the best success with the LINC of any of them. On the other hand John
Lilly who was interested in using the
machine to communicate with dolphins in a very wet
laboratory in Florida probably found the computer more
like the W. Gray Walter model. He would say that the
computer was nothing more than a status symbol with a
conveniently flat top.
It of course wasn't all just computers. Henry Littleboy
and Dr. Bernie Weiss, one of our visitors from the
Johns Hopkins, planed during lunch hour and of course
during all this time we were debugging and figuring
out what modes had simultaneous that could be put into
each of these six or eight machines. The evaluation
program formally funded 16 of them but there was something
like 21 of them before we were through.
People came in to sort of get in on the act. We had to
keep them modified as we discovered in our debugging
as we went. We were really making half a dozen or ten
simultaneous prototypes. Finally it was time to
plug
in the DEC plug-in units. the system units which
went in sideways into the frame. Here is Dr.
Benow<?> Duke University putting in some
modules. As you can recall some of those old units had small
holes in the bail and a screwdriver had to be
inserted from time to time to adjust the pot on some of them.
The sense amplifier was one of them that was a
nasty one. It was bit of a trick reaching into the frame and
turning the screwdriver sideways and getting
through into the hole in the bail and finding the slot of the
screw and turning that darn thing while it's
<?> on the scope. Everyone managed to do that successfully
except one chap who was a
brain surgeon.
We wired up the cables. This is a picture of Dr. Grodinson
<?> Randell stringing together some of those long
cables that tied the electronics frame to the rack.
Here is a number of assembly console shots of things
mounted in racks and so on. That is Norm Hill and Ralph
Stacy. Stacy and Waxman have published a four
volume series in which this is volume two 'Computers in
Bio-Medical Research' in which they have
summarized a treat deal of the work over that decade.
Here is a picture of Norm Kinch who is here in the audience
as you know. He is stacking up some of those
console modules. You can see that there are four of them
arranged to be stacked. Those are wood boxes;
they are heavier than they ought to have been, but a lot
lighter than you need a strong man in order to handle.
Norm in fact did everything as Dick said in his introduction.
Norm came to us at Lincoln Laboratory when
we were starting the TX-2. He immediately took charge of
all there was to do and I must tell you that the
way to make sure of the success of a project, if at all
possible, is to try to find as talented a person who
happens also to be an ex-chief petty officer. This guy
made everything happen. Norm if you will forgive me;
he is the sort of fellow to whom you could say I would
like to have a laboratory built on the other side of
town and stocked by Thursday afternoon and Norm would
say okay boy I'll take care of it. He generally did.
This was a very intensive period and we did set very
tired. Some of these pictures speak for themselves. I
believe that is 3:30 a +m. Finally the machines were
finished and shipped off by a van to their destinations.
We had done a neat thing and we deserved a big rest
and a big pat on the head. As it happened the center
for computer research in the bio-medical sciences
fell apart at that very point. In retrospect it was too much
too soon for MIT to absorb properly and conditions
were laid down and they were simply impossible for us
to meet.
Without being too detailed about it or too morbid,
in fact what happened was we sent the money back to NIH
and once again we are on the road looking for a
place to continue the work. NIH was very interested, of
course in seeing us continue seeing we had thus
caught them at the end of phase one of a two to three
phase program. Furthermore, we had other tricks
up our sleeves, though we didn't describe them very well. I
had in mind the macromodule stuff which we were
hoping to build; a new effort but in any case it wasn't
possible to do that at MIT so for the next many
months we tried to find another place to live.
It isn't so that you can just take an arbitrary
computers look into its history and say well somebody decided to
do that and sat down and did it and out came a
computers nice jobs go on and do something else. In fact you
see nothing to have been farther from the truth
in the case of the LINC. It had a very stormy history. The
LINC group was a very tightly knit and dedicated, talented
bunch of people who constituted a kind of an
inadjustable lump for MIT. They just weren't able to rationalize
our position with respect to the traditional
and right correct position for the University to take. It is very hard
for research of this kind to be done at the University auspices anyway.
Well, so we moved. We went to St. Louis where this
provost Dr. George Pake now vice president of the
Xerox Corp. had warmly welcomed us and even guaranteeing
to put up the money in transition should
we not be able to find enough to sustain the operation
during that period until new funds could be
secured.
That is a picture of the exterior of the old Shriner's
Hospital for crippled children which was partially
occupied and partially renovated at the time and in
which Jerry Cox had his repertoire with whom we had
had many dealings. It needed a little work but we did in fact fix: it
up.
That is not the only space but that was
one very dramatic one. Most of the building did look
like that. It had been abandoned for many years and St.
Louis is a very muggy climate. Here you see us dressed
up and we did install a LINC, that is a very clean
installation. Dr. Russ Pfeiffer was from MIT and he
built up with Charlie a laboratory for auditory
neurophysiology at the Medical School which the
Shriner's Hospital was adjacent to.
You can see Severil<?> and Jerry Cox not holding
us to the console, but operating the switches for some of
the displays that you can't see in the foreground.
Mort and Win Hindle talking to Bill Papian. Of course there
were side demonstrations as well. I think Dr. Maynard
<?> of the St. Louis group demonstrating really great
program that showed massive membrane motion. There
is a picture of me talking with Dick Clayton and
Win, and Mort talking with Professor Harold Shipton
whom some of you have seen here at DEC from time
to time.
From Duke University I'm very fond of this one.
'Astonishment is the only way I can describe me feelings
about the performance of the LINC except for a
few tape reading difficulties which were adjusted, the
machine has given no trouble whatsoever. It
has been completely reliable as far I knows, and in 4000 hours it
has never made an error which wasn't attributable
to something else when we finally understood what the
difficulty was. On several occasions we have had
the machine running non-stop for as many as three weeks
with no sign of trouble. If only the mechanical
equipment were as reliable my maintenance troubles would be
over.' That is a familiar story. Gus I heard
you make that comment in 1952.
Here is one from Stanford. 'The performance of
the LINC in respect to maintenance had reasonable
expectations. After approximately 3200 hours
of operation the only failures have been one bad cable
connection and two output transistors whose
failure can be traced to external misuse.' from Presbyterian
Hospital 'we have been most satisfied with the
LINC in our lab. The computer has practically been in daily
use and has permitted us to execute a large
number of experimental work which would not have been
possible otherwise. People had trouble
programming and that is no surprise.'
DCM001.4 |
Return to On-Line Documents