Outside Entrance
Visitors at the opening.
Return to List of Reports
Highlights from
Volume 11 ---- Winter 1984/85 |
Contents of Highlights
|
Outside Entrance |
Visitors at the opening. |
The re-opening and re-birth of The Computer Museum took a long time in the making. Marlboro provided an excellent beta-test site for historic exhibits but gave us little experience about interactive computing within exhibits.
After the Board of Directors approved the move in May 1983, planning started immediately. A team of "developers" was put together. Dr. Oliver Strimpel, then Curator of Mathematics, Computing, and Navigation at The Science Museum, London, agreed to come as Visiting Curator and develop a highly interactive gallery devoted to computer graphics and image processing. At the completion of this work, Oliver agreed to stay on as the Curator of the Museum. Oliver subdivided the tasks in the image gallery with Geoffrey Dutton and Andrew Kristoffy as developers.
I undertook the role of curator of the rest of the exhibitions with "developers" for each segment: Paul Ceruzzi (who is now at the Air and Space Museum) on the 1950-69 Timeline; Beth Parkhurst on the integrated circuit and Apollo Guidance Computer exhibits; Carl Sprague on the "See It Then Theatre"; Meredith Stelling on the ANFS/Q7, SAGE, and UNIVAC exhibits; Gregory Welch on the IBM 1401 Room, Seymour Cray, and Manufacturing exhibits; and Bill Wisheart on the personal computer exhibit.
Oliver, the developers and I then started to work with a broad set of advisors who helped us refine ideas, collect the materials and computers, and some of whom eventually worked on the actual programs and installations. The architectural firm of Crissman and Soloman were chosen to integrate the ideas of the developers with the existing structure of the 1880's wool warehouse and come up with suitable exhibition space. Meredith Stelling took on the role of supervising the contractors, Hawkins and Co., and the graphics designers, Maxwell Design.
When we worked out the schedule, all planning was to be complete by June 1, construction complete in early October, with a month for exhibit installation. It never worked that way. Everything happened at the end. And is still happening. When we opened with over half an acre of exhibits in five large rooms, each was about 70% complete. Over the winter, the exhibits will be finished and some will start to evolve even further as we watch how visitors are reacting.
By June 1, the developers had their scripts completed and then seriously sought to implement them. One exhibit that we knew we wanted to animate was on the Apollo Guidance Computer. Hewlett-Packard agreed to give us an HP-150 with a touch sensitive screen and the use of Tom Horth in their Andover facility as a consultant. Draper Laboratory's Malcolm Johnston coordinated the work of our summer intern, Andy Gerber, in order to ac- curately simulate the astronaut's console. But by July 1, the HP-150 had not appeared. Andy was more than ready to get started on the machine. Tom Horth came up with a loaner so that the project could begin in earnest. By mid-August the prototype program was tested and it was slow. Tom arranged to get us a faster compiler. Then, the actual machine came in September after Andy had gone back to MIT
Another interactive exhibit that we wanted from the outset was one that communicated the concept of "discernability," conveying the meaning of pixel sizes, grey levels, and false coloring in image processing. Masscomp agreed to take on this exhibit. Lorrin Gale, Vice President of Engineering, personally made two trips to the Museum with several programmers. The project was specified and Masscomp produced a special two terminal machine. Each terminal was connected to a tv camera that they supplied. One camera is focussed on the face of the visitor, who then can change the pixel size and grey levels of his own image. The other camera is focussed on the view of Boston. The visitor can then color in the grey levels to create an "Andy Warhol-like painting." The engineers at Masscomp got excited about this project (one that has little hope of ever being a product) and kept assuring us that it would be exactly what we specified. Oliver visited it at the plant three days before opening and was satisfied. Masscomp delivered the two exhibits exactly one hour before the preview for the Board of Directors!
Last July, Oliver, Geoff Dutton and I went to SIGGRAPH, where, among other things, we collected "the teapot" from Martin Newell and got lines on other exhibit material. As I write this on New Year's Day the "teapot" exhibit is not yet complete. Its components are numerous. Adage gave us a terminal connected via a fiber-optic cable, donated by Fibronics, to the VAX 750 contributed by Digital Equipment Corporation. The "teapot" simulation is still being programmed by Allan Sadoski, a volunteer from the Adage user group, and his 16-year old "hacker friend" Neil Day. They are spending most weekends at the Museum, providing a living, working exhibit. Parallel to this simulation, the Design and Production staff of The Children's Museum is building a stage set for the real teapot where its lighting can be manipulated manually. This should be complete in mid-winter.
IBM Fellow and Harvard Professor Benoit Mandelbrot became very excited about producing an interactive exhibit of his concept of fractals. He produced a program on the IBM XT but it lacked sufficient variation. A prolific author, he discovered, as we had, that an interactive exhibit needs to have a lot more variety than the illustrations within an article. A week prior to opening, the program was finally acceptable but we had no machine to run it on. Our two IBM XTs were committed to other programs. Dr. Mandelbrot arranged for another XT for this exhibit and it arrived (minus several critical parts) three days before the opening.
One exhibit that arrived complete and wonderful a full week before opening was a video of the view done by Dean Winkler and John Sanborn of VCA Teletronics. In August, they came up from New York and cavorted on top of the roof videotaping the view. They talked to us, looked at the logo and some of our concepts, and then spent over 200 midnight hours of editing with the very fancy frame-buffering equipment to produce a three-minute spectacular of the view popping out in different colors with the core plane logo flying over it and skyline circling a pyramid. In this case, the creators were given artistic freedom and went wild in making a very spectacular video. The equivalent spot made commercially would cost hundreds of thousands of dollars. Dean Winkler and John Sanborn will come up and explain to all how this was done in a talk on Sunday March 17.
Yes, it's great to be open. Three "beta-test" talks were given in December, and now the full schedule of talks for the spring appears on the inside back cover. These are planned for every Thursday night at 7 and Sunday at 4 from February 7 to April 28. The next issue of the Report will have an article on one of the December talks-a conversation between Steve Levy and some of the heroes featured in his book Hackers. For those of you who can't get to the talks, we'll try to bring you the very best in the Report.
Best wishes for the New Year.
Whirlwind Entrance
Exhibits
Diagram
| Entrance into the Museum puts the visitor in Whirlwind's arithmetic units, which occupied a whole room in the Barta Building at MIT. The 16 bit word length, extending 32 feet, was partially determined by the width of the room. |
|
The visitor enters into the Whirlwind computer-the first real-time stored pro- gram computer, so large that it took up a whole building. In a segment from a 1951 "See It Now" program, Edward R. Murrow interviews "the Whirlwind electronic computer". After he has Admiral Bolster give the "whirlwind its workout," Murrow says, "Well, I didn't understand the answer, and I didn't even understand the question." This seems really quaint to today's visitor because the whole program that the Admiral wants run on the building full of Whirlwind, is running on a Compaq that was programmed by summer student.
This first exhibit illustrates the revolution, the unbelievable power of the first computers in the early fifties, and their incredible evolution in thirty-five years. The Whirlwind occupied a building, consumed 150 kilowatts and cost as much as $20 million. The equivalent personal computer sits on a desk, plug into a wall socket, and costs two thousand dollars.
The AN/FSQ-7 and SAGE System
The Q7, a production version of Whirlwind, was probably the largest and
longest lived computer
in existence. It illustrates the computer components that are now on a
single board or micro-chip.
The arithmetic and memory units with their 55,000 vacuum tubes took a very
large space. The
visitor can walk through the seven foot high banks of
vacuum tubes and up to the four foot by four foot by eight foot 32-K core
memory stack. The
equivalent chips are exhibited and a terminal to the VAX provides a
tutorial on how core
memory works.
The control consoles were so large that they took up an entire room with
several operators. The
activities of the other components of the machine were shown in flashing
lights on the consoles
and the operator had a telephone to communicate with the people on the
arithmetic, input-output units, or generator for the power.
The "Blue Room" consoles had large round screens that showed aircraft
moving across the
airspace. The screens were updated every 15 seconds by the Q7 causing a
constant irritating
flicker,
hence a soft blue light in the room for the purpose of seeing the screen.
The consoles display the
air situation display and some were especially designed for weapons
assignment or
interception. The exhibit includes the consoles, chairs with their special
drawers on the seats,
and ceiling panels to recreate the feeling in the "Blue Room".
A console from the SAGE Blue Room, the control room for the SAGE, the U.S.
air defense
system from 1958-1983. Here, Computer Museum visitors can see the oversized
video
display terminals that served as the first computer graphics output devices
that used light
guns to identify the airplanes shown moving across the screen.
SAGE Blue Room.
Visitors walking through two rows of the AN/FSQ-7 arithmetic unit. Each
computer had
55,000 vacuum tubes with 300 changed each week for preventive maintenance,
whether
they needed it or not.
After UNIVAC I was featured predicting the Eisenhower election of 1952,
the name almost became synonymous with "computer." The video-tape
and components of a UNIVAC I bring this era back to life.
J. Presper Eckert, Walter Cronkite and Charles Collingwood with the
UNIVAC on election night in 1952. At 8:30 p.m., with only a few million
votes tabulated, UNIVAC's first prediction showed a landslide victory
for Eisenhower. Since nationwide polls had indicated a close race,
Remington Rand officials revised the national trend factor and had
UNIVAC recompute. At 9:15 p.m., UNIVAC publicly predicted 8 to 7 odds
for Eisenhower. By 10:32 p.m., all predictions showed that Eisenhower
would decisively beat Stevenson (442 to 89 electoral votes). The president
of Remington Rand went on the air to explain why they had tampered with
the original prediction.
The first two generations of computing are illustrated in a timeline with
artifacts that move the visitor year-by-year over this twenty-year span.
The invention of the transistor is at the beginning and the introduction of
the NOVA, a third generation integrated circuit computer at the end.
Unique artifacts, such as a unit from the EDSAC and the ILLIAC I, are
complemented with illustrations of new technologies, applications, and
ephermal materials such as "Do not spindle" buttons.
The timeline is meant to be evocative of a walk through history. We hope
that it will also bring to light many hitherto buried artifacts for
preservation
as part of the history of information processing.
Gordon Bell and Mass. Secretary of Commerce Evelyn Murphy looking at
the early sixties section of the ."Timeline." A module from the ILLIAC 2
hangs over an Olivetti Programma next to the teletype. Over 100 artifacts
are included in this twenty-year timeline.
This picture of the 1969 Data General Nova and three of the company's
founders, Edson de Castro, Herbert Richman, and Henry Burkhardt, ends
the Timeline.
UNIVAC I
Computing from 1950-1969:
A Year by Year Timeline
| Seymour Cray and John Rollwagon, President and Chairman of Cray Research, stand next to a prototype of the CRAY-2. To keep the components cool, the entire CPU will be immersed in inert fluorocarbon, the substance used for artificial blood. | 
|
In 1972 Cray left Control Data to form his own company: Cray Research Incorporated. After fours years of work, Cray Research delivered the Cray 1 to the Los Alamos National Laboratories in early March, 1976. Its radical design and $8 million price tag led some to call it "the world's most expensive loveseat." A section of the Cray 1 is on exhibit at the Museum. Above it is a large image of the computer which was generated by a Cray 1 computer, illustrating the use of the large computers for graphics and entertainment applications as well as the large-scale number crunching.
| Associate Director and Curator of The Computer Museum, Dr. Oliver B.R. Strimpel, and Harvard University professor, Dr. Benoit B. Mandelbrot, also an IBM Fellow at the Thomas J. Watson Research Center, are shown standing with "Fractal Planetrise," an artificial computer generated landscape in "The Computer and the Image," a major gallery at The Computer Museum. Fractals are mathematical objects developed by Dr. Mandelbrot and have been used as models of natural phenomena such as,' turbulent fluid flow and the shapes of rivers and coastlines. Fractals have recently played a role in the synthesis of artificial landscapes for the film industry. | 
|
Several exhibits use the fine view of downtown Boston from the gallery indow as a starting point: a television camera captures an image for the visitor to color in digitally, a plotter continuously draws differently colored and shaded views, and a video shows both a walk through a 3-dimensional database of the city as well as an exhilarating range of special effects applied to stretch a 2-dimensional version of the view into " 2.5" dimensions.
The techniques of realistic image synthesis are shown in the section, Building an Image. Lighting, subtle color shading, the simulation of texture, transparency, reflections, and refractions of light are all shown. For many years, researchers in computer graphic realism used the data set that graphically reproduced Martin Newell's teapot to test their methods. The original teapot is now on show here in a mini stage set, next to a computer generated rendering of itself, complete with artificial colored lights. Here too you can browse through 3-dimensional computer models of houses on offer by a commercial builder.
A section on computer-aided design lows images and objects designed with the help of a machine. Examples range from parts of a Boeing 757 to an Olympic running shoe. At interactive stations visitors can design a car and complete the design of an electrical circuit. A large high precision pen plotter draws the artwork required to fabricate a microprocessor chip.
Interactive demonstrations allow the visitor to make his/her own fractals and cellular automata. Both are useful models of some natural phenomena, and rely on computer graphics for their investigation. Fractals are useful in generating artificial landscapes, several of which are shown here. In a section entitled Simulation, a video shows examples from the modelling of galaxy collisions to the interaction of a DNA molecule with a drug. The fantasy world of SPACEWAR!, the first computer game written by MIT hackers on the DEC PDP-1 computer in 1962, is demonstrated on special occasions on the PDP-1, and otherwise runs on a modern micro. Visitors can also fly a Cessna using a flight simulation program. A video shows state-of-the-art use of graphics in flight simulation, landscape synthesis, education and advertising.
Perhaps the most appealing use of computer graphics is in the making of films, both for animation and for the creation of convincing fictitious scenes. A computer animation theater shows a series of films from the earliest use of key frame inbetweening to the latest offering from Lucasfilm, completed in August 1984.
The visitor should be able to sense the excitement and challenges of this rapidly changing field in computer applications, as well as absorb many of its fundamental concepts. Much of the film, video material and working demonstrations will be updated to keep abreast of developments.
by Robert N. Noyce
As I was driving in tonight, I was listening to a Chrysler ad
pointing out that the company was 60 years old. I think of
Chrysler and the auto industry as old. Then, I thought, the
semiconductor business must be reaching middle age, since it
is now over 30.
In 1954, the semiconductor business
amounted to 25 million dollars, the growth sequence then was
35, 80, 140, 210, 360, and then 550 million by 1960. Half the
business was in transistors; silicon accounted for a relatively
small share.
In the fifties, everyone was trying to figure out new and better
ways of making transistors. At one of the solid state circuits
conferences, an explorers kit, designed to keep you from
getting lost in the woods, was displayed. It consisted of a box
with a small cube of germanium and three pieces of wire. If you
got lost, you were to start making a point contact transistor.
Whereupon ten people would lean over your shoulder and say
"That's not the way to do it." Then, you would turn around and
ask, "Where am I?"
At the time, germanium alloy transistors were made by putting
indium on top of semiconductor germanium and melting it just enough to
dissolve some of the germanium and then recrystalizing it on
both sides to make a PNP transistor.
One baffling research question was why germanium, when it
was heated and then cooled in the laboratory, changed from N
to P type. Simultaneously transistors were being manufactured
with N type germanium on the factory because the indium acted
as a getter to pick up all the impurities instead of converting the
germanium.
In the mid-fifties, the thinnest possible transistor was a fraction
of a mil and a mil was a megacycle so these weren't very useful
for anything except for hearing aids.
Between '54 and '55, we started worrying about diffusion as a
way of getting impurities into the semiconductors, giving good
control of the depth dimension. The problem was to get control
of the other dimensions. Some of the first work was done at
Philco because the semiconductor group worked right across
the hall from the laboratory that was working on etching
shadow mask tubes for color television. They were experiences'
with photo engraving, which turned out to work a lot better.
The invention of the planar transistor by Jean Hoerni further
set the stage for the birth of the integrated circuit. Planar
transistors solved the problem of impurities on the surface of
the transistors and at their junctions that had been lousing up
the specified characteristics. Hoerni's idea was to leave the
silicon dioxide, a very good insulator, on top of the transistor
when it was being diffused, thus forming a protective cover.
The government gave further impetus by their interest in
getting things into smaller packages. The Air Force project
Tinker Toy and the concept of molecular engineering didn't
really work very well, but it did let everyone know that there
was an interest in getting things small. A square inch chip with
ten thousand transistors was very labor intensive: each
transistor had to be attached by a couple of wires and soldered
down. There had to be a smarter way.
I remembered that when I was in college, I could slave over
something, finally get the right answer, hand in my paper and it
would come back with big red markings on it. My physics
professor would say I did it the hard way. Then he'd jot
down a couple of sentences which clearly made
it much easier for me by using some other
method. I guess that is what stuck with me
because one of the characteristics of an inventor
is that he is lazy and doesn't like to do it the hard
way. Putting those 20,000 wires on 10,000 chips
of silicon seemed like the hard way to me.
Although the printed circuit board was starting
to be used, the thought of printing a circuit on
top of the transistors had not occurred. It was
the genesis of the idea of the integrated circuit.
All the elements were converging: photo
engraving enabled reproduction and the planar
transistor allowed conductors directly on top of
it. Three ideas popped up at that time. One was
junction isolation, which I patented, even
though it turned out that Kurt Lehovic had
thought of it years before at Sprague. J. Last at
Fairchild thought of the idea to etch the
transistors apart, glue them down to something
and if you still knew where they were you
hopefully put them together. This idea had been
previously patented at Bell Labs. The one I did
get a patent on used intrinsic isolation, that is to
use the silicon as an insulator. It didn't work well
at first because by bombarding it with neutrons
or doping it, leakage occurred and the life was
too short. Junction isolation is now being
broadly used.
After the original concept was developed, things
moved very slowly. One reason was the low
yield on transistors: with 50% yield and ten
transistors together, the final yield of one over
two to the tenth is a small number. We didn't
even consider putting a thousand transistors
together. Another problem was that the early
integrated circuits were very slow. And, of
course, the market was opposed to this
innovation.
Progress followed the classic Moore's curve.
Every year you could get something twice as
complex as the year before. That extrapolates to
a million elements in 1980. We didn't quite make
that unless you allow for the introduction of new
things like magnetic bubbles. The technology
also changed from bi-polar to MOS.
Costs are determined by complexity and the
number of leads per square inch of silicon with
problems setting to 20,000. Starting with a 5/8th
inch wafer in 1963, costs were reduced by
increasing the size to 1.5 inch in '65 and two
inches in 1970. The die size and area were also increased
to reduce the density of defects that would kill the surface. It became
possible to use an ever increasing area to put a
circuit on and have it work. Circuit dimensions
themselves have been reduced below the size of
neurons, 10 microns, and these are being used
for speech synthesizers and other products.
Today, we have two micron circuits ands are
talking about .7 microns, so we
indeed are getting down to biological
dimensions and it is conceivable to talk about
things the brain can do.
Other new ideas were important. One was MOS
and the second was epitaxy. Prior to the use of
epitaxy only the surface could be more impure
than the underlying material. This was another
bag of tricks.
The first set of integrated circuits had straight
Boolean functions. With progress the designers
wanted complexity with lots of leads out of a
circuit and the semiconductor manufacturers just
didn't like that at all. In addition, the more
complex products had a lower demand, and as
manufacturers we were thinking of making
millions of items Simultaneously the computer
companies in the early seventies were talking about
tens of thousands per year. One kind of chip,
however, was like heroin to the computer
designers and that was memory Give them a little
bit and they want more. Thus, memory chips
became a major standard product.
What has the chip wrought?
The chip has been one of the main elements
allowing the ubiquity of computers. Computers,
as tools and devices to help train people to think
logically and work precisely, have caused a
major revolution in education, business,
government, and all aspects of society. The
telecommunications manufacturers would have
us believe that every telephone in the world will
be a computer terminal.
Some people fear this idea, just as I feared the
telephone. One day when I was quite young, my
folks were out and left me alone. The telephone
rang. I panicked, picked it up, and said, "Hello,
nobody's home." Then hung it up. Today I
can't imagine living without a telephone.
Let me point out a couple of other changes that
I've observed. The first computer in an
automobile only controlled the non-skid brake
and exhaust
and it cost twice as much as the car and filled
the whole trunk. In fact, the rear seat had to be
used as well in order to install the computer.
Today computers in cars do ten times more
work and cost about $30. They are less expensive than a
mechanical carburetor and will pay for itself in
the first year in gas savings.
Jobs in the future are not going to require the
skills of the past. Onehundred-and-fifty years
ago, 50% of the American labor force was
employed on the farm. Fifty years ago the
greatest proportion was in manufacturing.
Today that is about 20%. These latest
statistics are inaccurate because the
categories have not changed with the
economy. Intel is included in the
manufacturing sector, even though only 30%
of our people actually touch any products that
are shipped. Most of our employees sell, keep
books, or even do such useful work as design
the next generation of products. Today more
than 50% of the labor force is working with
information.
The computer is the major tool that can help
information workers. It's a productivity
enhancer for people who work with ideas as
well as for people who work with things. It will
allow more human use of human beings. Dull
repetitive tasks are the first to go.
example, retyping a letter for one
mistake, or reformatting a marketing forecast.
The tradition of liberal arts education was
designed to allow people to understand and
communicate in society.
Grammar, rhetoric and logic came first, and
then the quantitative studies of arithmetic,
music with its geometrical relationships,
geometry and astronomy followed. The same
task is essential today. The student has new
tools to help understand the continuing
accelerating advances in technology. Most
students will be working with a computer in
some way
It's not necessary for society to breakdown
into C. P Snow's two cultures in which those
who do not work with technology are left
behind those who have the modern tools to
become productive. Despite the advances in
technology, math, science and engineering are
not attracting enough people in the US. The
power of our computers that can help people
as tools is growing beyond common
imagination.
The Computer Museum has the CDC 6600, the
first production supercomputer from 1963. It
cost more than $3 million and only had 500,000
transistors. That will be available on a single
chip within a couple of years and everyone
can have a supercomputer. All the educational
institutions have a challenge to make this work
for the science and liberal arts.
The microprocessor or microcomputer was
introduced by Intel in 1971.
Return to List of Reports
The Integrated Circuit:
Origins and Impacts