The First Computers

1
The First Computers Electronic Prototypes
I ENIAC and the the Firing Table Crisis
2
The scientific study of the dynamics of
projectiles, as bullets -- that is the art of
ballistics, until the total mechanization of
warfare in the 20th century, was far from being
an exact science. In a series of experiments
between 1864 and 1880 the Reverend Francis
Bashforth related air drag to velocity and
produced the first firing tables to be consulted
in the battlefield as a guide to the ranging of
artillery pieces. These tables were refined by a
French military commission but at the onset of
the First World War, there was still a lot of
guess work involved.
3
Many factors were simply not calculated. Since,
without firing tables, any artillery weapon is
much reduced in utility, the Americans took
ballistics very seriously and established a
tradition of research I the aftermath of that
war. At the Aberdeen Proving Ground a succession
of highly educated ballistics officers appeared
during the 1920s and 1930s and it was these
officials who moved to acquire a differential
analyzer.
4
Nevertheless, the ballistics work generated by
the Second World War was overwhelming. Each
firing table for every new weapon required the
tabulation of dozens of factors across thousands
of possible trajectories, any one of which
represented half a days work for a human
computer with a desk calculator. By the summer
of 1944 the computing branch of the Ballistics
Research Laboratory was producing fifteen tables
a week, but the need was for forty. This, the
so-called firing table crisis, was the
supervening necessity for a fully electronic
complex difference analyzer, the machine which
became known as ENIAC.
5
The machine not only eliminated delays, but it
also permitted far more extensive ballistic
calculations.
6
Unlike the electrical mechanical calculating
machines that used general-purpose
electromagnetic relays of telephone quality, such
as the Mark 1 and the Zuse machines, John
Atanasoff, a professor of physics at the State
College of Iowa, introduced a different sort of
binary machine. He coined the term analogue
computer to describe the semi-mechanical complex
differential analyzers. Atanasoff was engaged in
building a binary digital device having about 300
valves. Its purpose was to solve linear
algebraic equations. Atanasoff, in addition to
using path-breaking valves in his logic circuits,
used capacitors arranged in drums as the store
and a built-in pulse by which the machine could
internally time its own operations.
7
In December 1940 Atanasoff read a paper on his
work at a conference in Washington attended by
John Mauchly. In the 1930s, Mauchly had been
eager to avoid the slog of hand computations and
he claimed that he had built some experimental
circuits at that time using valves. Upon hearing
Atanasoffs presentation Mauchly went to see
Atanasoff to discuss in detail plans for a
general purpose electronic calculator utilizing
valves. Atanasoff never patented his valve
calsculator, was drafted on the outbreak of war
and never returned to computing machinery.
Mauchly continued the work derived from
Atanasoff. A judge in 1974 was to hold that
Mauchly got all his ideas from Atanasoff, yet,
confusingly, at the same time found that Mauchly
had invented ENIAC.
8
The basis of this opinion was that in August 1942
Mauchly had written a paper, The Use of High
Speed Vacuum Tube Devices for Calculating, which
led directly to the ENIAC project. Beyond speed
of calculation, he suggested greater accuracy ,
enhanced checkability, a certain degree of
programmability, easy faultfinding and labor
saving would result from the use of valve. At
first the proposal did not provoke much interest
and it took the discovery of the paper by a
Lieutenant Herman Goldstine nearly a year later
to get the funding and initiate the project for
Mauchly. It wasnt till May 31st in 1943 that a
request for 150,000 had been approved and work
began. In June the machine was renamed
Electronic Numerical Integrator and Computer.
Mauchly immediately involved his graduate
assitant J. Presper Eckert Jr. in the ENIAC
project, due to Eckerts knowledge in the use and
reliability of valves.
9
Just to clarify, when I use the term valves, I
mean a mechanical device that regulates the flow
of gases, liquids or loose materials by blocking
and uncovering openings. And the big gamble in
ENIACs design, then, was in the number of valves
required. The original plan called for 17,000
tubes operating at 100,000 pulses a second. It
had theoretically 1.7 billion opportunities to
break down every second it was switched on. But
Eckert, understood that, if valves were left on
constantly, the failure rate was greatly reduced.
If Eckert were to be proved right -- and he was
-- ENIAC would also domonstrate that there was no
technological reason for an electronic calculator
not to have been built much earlier.
10
Eckert and Mauchly began work on the ENIAC a week
before the final contracts between the Moore
School, the university where Mauchly taught and
Eckert studied, and the Ballistics Research
Laboratory were signed in June 1943. ENIAC was
to take three years of painstaking labor and
there were various reasons. Although the work
was tricky it could be done. Getting the parts
off the shelf, in the face of wartime shortages,
was much more difficult. It is likely that, had
the Ballistics Research Laboratory set out to
have an ENIAC built in the 1930s, it would have
been finished more quickly than in the mid-40s.
Weeks even months were wasted waiting for
electronic supplies. The racks were made by a
kitchen cabinet manufacturer in New Jersey who
was going out of business for lack of steel.
11
By November 1945, ENIACs debugging procedure
started and at von Neumanns suggestion, a team
from the Manhattan project at Los Alamos went to
Philadelphia with a problem for ENIAC to
calculate. Early in the New Year, the Moore
School and the Army elaborately prepared to
unveil ENIAC to the world. It had eventually
cost 500,000, but then so had the latest Bell
Mark 5 electro-mechanical device, and although
the war was over, ENIAC did what it was designed
to do We chose a trajectory of a shell that
took 30 seconds to go from the gun to the target.
Remember that girls could compute this in three
days, and differential analyzer could do it in 30
minutes. The ENIAC calculated this 30-second
trajectory in just 20 seconds, faster than the
shell itself could fly.
12
There was already a degree of awareness, from the
Robot Brain ceremonies for the unveiling of
Aitkens Harvard Mark 1, that a new generation of
super devices was emerging, and on February 15th,
1946, upon the public demonstration of ENIACs
powers, the New York Times quoted leaders as
heralding it as a tool with which to rebuild
scientific affairs on a new footing. Yet
confusion in the popular account between the
earliest computers and these powerful electronic
calculators which immediately preceded them
existed because the line between number-crunchers
and symbol-manipulators was in the 1940s a thin
one. A symbol-manipulator, as opposed to a
number-cruncher, must have a built-in set of
instructions which can be varied by the operation
of the machine itself as a result of its actual
calculations and it must therefore have an
extensive data-store -- and this is a computer.
13
Since ENIACs progress was so slow, Eckert and
Mauchly, began to generate ideas for
improvements. Improvements primarily having to
do with memory and programmability of the
machine. They wished to use teletype tape for
memory, so that tapes could be cut for many
problems and reused when needed. The program
here is still being envisioned as elementary
instructions. By August 1944, Goldstine began
lobbying for a new machine to utilize the
advances being suggested. William Shockley, a
Bell Labs scientist, had developed an elegant
device that allowed electronic pulses to be
converted into a physical form and then
reconverted into electrons.
14
By January 1944, using this development, Eckert
had outlined a memory system for the calculator
in which mercury-filled tube, 4 and a half feet
long would be arranged so that the pulses
(representing the digits in the calculator) would
activate a quartz crystal at one end of the tube.
This would create an ultrasonic wave in the
mercury. The wave would travel infinitely more
slowly than the pulse. The slow-down would act
like a memory, storing the pulse. At the tubes
far end, another crystal would convert the
ultrasonic wave back into a pulse. Place the
mercury tube in a loop and a stream of pulses,
digits, could be made to circulate continuously
until needed by the computer. The machine would
remember the digits.
15
A number of people proposed a similar read-write
storage system, so it is uncertain of who came up
with it first. Von Neumann was a major figure
who begun visiting the ENIAC team in 1944. At a
series of meetings he and the team explored the
problems of logical control that the storage
lines would create. Von Neumann organized the
memoranda of these meetings into a report, issued
on June 30th, 1945, First Draft of a Report on
the EDVAC. EDVAC stands for Electronic Discrete
Variable Automatic Computer, where the crucial
term is Variable. The first report was published
at the university where Von Neumann taught and
failed to credit the others. This paper
publicized the stored program concept and was the
earliest comprehensive written outline for the
modern computer.
16
The war constituted the supervening necessity for
the advanced electronic calculator and, at its
end, all the teams involved in such work were in
danger of being returned to other projects. The
firing table crisis was not enough to maintain
them and general scientific or business
requirements were no more visible in 1945/6 than
they had been through the 30s. However, the idea
of the stored program machine -- the true
computer -- was not lost as a new necessity came
into play during the first years of peace.
17
ENIAC was finally working by November 1945.
Nicholas Metropolis and some colleagues from Los
Alamos had already the previous summer, at John
von Neumanns suggestion, come to Philadelphia to
explore how the machine might aid the computation
of some problems in thermonuclear ignition. In
November, Metropolis returned and ran the
calculation. Scientist from Los Alamos were so
early on the machine that their problems were
used to debug the device. The results appeared
on punch-cards containing no indication of what
they were or what they were intended to
represent. The study of the implosion problem
gave one of the great impulses to the development
of fast computersbecause it had to solve these
problems Los Alamos, consciously or not, made a
great and fundamental contribution to the
development of computing.
18
One war was over, but another, with a nuclear
arms race at its heart, was starting. There was
a slow down in computer development as the arms
race was not immediately realized, the Russians
would not explode an A-Bomb for four years.
Aside from the academy, elsewhere in America,
computer plans were being abandoned. Mauchly and
Eckert having left the Moore School and gone into
business for themselves, nearly went bankrupt
trying to sell computers. Even at Los Alamos
there was a hiatus. Nicholas Metropolis had left
for the University of Chicago in 1946. But then,
two years later, he was called back and began to
build his computer along lines laid down by von
Neumann who was back at Princeton.
19
Whirlwind, the path-breaking real-time machine
that Jay Forrester of MIT was building for the
Office of Naval Research to study sircraft
stability design, was on the point of closure
when it was realized it would serve to meet the
perceived threat of Soviet air attack. In August
1949, the Russians exploded an atomic bomb and
the specter of a Soviet nuclear strike over the
North Pole suddenly loomed. A Whirlwind member
recalled
20
A major threat came from the low-flying Soviet
aircraft. At low altitude, radar range is very
short, and information from many radars had to be
netted to cover large areas. George Valley at
MIT was concerned about radar coverage. He met
Jay and it turned out that what he needed was a
real time digital computer and what we needed was
air defense. So Air Force financial support
appeared in the nick of time.
21
In June 1950 the Korean War began. By November
instead of being abandoned Whirlwind had acquired
airforce backing a 175 personnel. The first
hydrogen bombs were exploded in November 1952
(US) and August 1953 (USSR). The perceived
threat of Russian nuclear capability was
apparent. IBMs proto-computer, the
Card-Programmed electronic Calculator (CPC),
built in essence out of available IBM bits was
commissioned by Northrop as part of guided
missile development and was delivered in six
weeks. Thomas Watson, IBMs president was now in
the defense business. He telegraphed President
Truman offering IBMs full service to the
government.
22
The first IBM computer proper, the 701 was
originally designated a defense calculator and
the first customer was Los Alamos. Of the next
seventeen 701s, nine went to aircraft companies,
two to the Navy, one to the Livermore Atomic
Research Laboratory, one to the Weather Bureau
and one to the National Security Agency. The
first true computer to work in the US was the
Standard Eastern Automatic Computer (SEAC) in May
1950. It had such elements as a magnetic wire
cartridge from a commercial dictating machine as
well as valves and magnetic tape. SEAC was the
first computer to use solid state components --
(10,000) diodes -- for its logic circuits, (750)
reserving valves for amplifying functions. It
remained operational well into the age of the
transistor, finally being closed down in 1964.
23
A number of other computers followed, built by
various factions and making different
contributions. One major contribution, in
January 1950 was the full scale Williams Memory
which reconfigured new valves to provide for the
random access of data. ORACLE used a 2-inch
magnetic tape system as a back-up to its
Williams Memory. Most rejected tape as a means
for memory because its access time was slow and
one could not depend on the quality due to dust,
stretching or other blemishes.
24
The full-scale MADM used a variant on tape, a
magnetic revolving drum store, as a back-up to
its Williams memory. Drums, introduced in
1947, were made of brass or bronze and either
nickel plated or sprayed with iron oxide and,
although still slow, were more accessible and
reliable than linear tape. Eventually Hay
Forrester of MIT -- who was working on the
Whirlwind -- would produce, in the final stages
in co-operation with IBM, a core memory drum
which by the mid-1960s became the industry
standard. (The naming fad of these early
computers was to name them with acronyms such as
ENIAC, EDVAC, MANIAC Metropolis, ORDVAC,
ILLIAC, ORACLE)
25
Whirlwind, amongst the most significant of these
early machines, had begun as an analogue
real-time calculator to build a sort of flight
simulator -- an airplane stability analyzer --
for the Special Devices Center of the Navy. It
was to be a cockpit and a computer and it would
fly like an airplane not yet built. The
computer part of the device went digital in 1945
when its designers, led by Forrester, became
aware of ENIAC and the other planned digital
machines. By 1947, the year Forrester began
developing the core memory, the design was ready.
It was working about four years later -- the
first device to use a visual display system and
(by 1953) the first to use the magnetic random
access core memory drum system.
26
Eventually whirlwind metamorphosed into the
central control for the Cape Cod air defense
system. This was the initial machine in the
entire Semiautomatic Ground Environment (SAGE)
Air Defense system. Whirlwind in its developed
form was the basis of an IBM series later in the
1950s and, although it was the biggest of the
early machines, it was, in its overall
architecture and design as well as in use, the
closest of them to the present generation of
personal computers.
27
1952 was the last year of the computers period
of invention. In November the US exploded an
H-Bomb. In the UK the worlds first two
commercial computers were already in production.
In January 1953 IBM shipped its first
production-line computer, the 701, to Los Alamos.
Eckert and Mauchlys commercially produced
UNIVAC, which had been used in television, with
an enormous flurry of publicity, to predict the
outcome of the 1952 presidential election. In
the public mind the computer age was in full
swing. And yet there were barely two dozen
machines world-wide. The machines were all
massive and unbelievably expensive, designed for
extremely complex work at the heart of the cold
war.