The team who built the Electronic Delay Storage Automatic Calculator (Edsac) at Cambridge University had been doing nothing so simple as tweaking a few lines of code: they had spent more than three years building the computer and learning to program it from scratch. Suddenly, after months of debugging, it worked: a teleprinter began printing out a table of squares. It was 6 May 1949, and you could see it as the beginning of the modern world. The Edsac kept going for nearly 10 years, but the man responsible for it is still going, rather more strongly than you would expect at the age of 89.
In 2003 Sir Maurice Wilkes drives daily to the Cambridge Computer Laboratory, where he has an office as emeritus professor, and his mind is as agile as ever. There are of course other people and other times with a claim to having given birth to computing. There was the Victorian engineer Charles Babbage, who described and tried to build programmable computing machines a century before the technology had matured enough to make them practical.
His work had been largely forgotten by 1937, when young Wilkes began work at the Computer Lab, then known as
the Mathematical Lab, under Sir John Lennard-Jones, a structural chemist who was interested in using machines to solve
differential equations.
A year earlier, Alan Turing had published his classic paper, On Computable Numbers, outlining a universal computing machine. But Wilkes recalls:
'It was purely theoretical. There is no hint in that paper of there being any practical applications. No hint whatsoever. I think he [Turing] made that connection afterwards.'
Computing at the lab involved mechanical desk calculators and analogue computers that solved mathematical problems by modelling them mechanically - just as an analogue clock 'computes' the time by modelling the rotation of the earth. There was nothing that is generally thought of as a computer today. 'No-one had thought of digital computers. That was all in the future,' said Wilkes. So, as it turned out, was his work on computing.
Wilkes, who had done his doctorate on radio waves in 1934, became a wartime boffin, pulled from the lab to work on radar even before hostilities started. When he returned to Cambridge after the Second World War, he found himself in charge of the laboratory. 'It was wonderful. I'd had six full years involved in radar and I got away as soon as I could and said to myself: "Let's do something constructive." There was a wonderful feeling of reconstruction in the world. Everyone felt this excitement of establishing peacetime values.'
Considerable steps had been made in computing during the war. A team under mathematician Max Newman built a proto-computer called Colossus at Bletchley in 1943 to help decrypt signals; Turing, of course, also worked there. Colossus could perform logical operations but no arithmetic, and the fact that Britain insisted on keeping it secret for 30 years limited its influence. Much more important to Wilkes was the 18,000-valve Electronic Numerical Integrator and Automatic Computer (Eniac) completed in 1945 in the US by a team led by Presper Eckert and John Mauchly, initially to perform artillery calculations. It was decimal rather than binary, it was very cumbersome to program, and the processor was not separated from the memory. but a lot of lessons were learned building it.
Out of the blue in 1946 Wilkes got an invitation to attend a series of lectures at Philadelphia's Moore School
involving the Eniac team leaders. 'It was very difficult to cross the Atlantic at the time. Shipping was very scarce. But I did get there for the latter part of the course and I got the whole story - all these new ideas. As far as I am concerned that is where they all came from,' Wilkes said. 'They were quite revolutionary. The idea that everything could be contained in memory: numbers, any mathematical tables that you needed, and indeed the program itself. All
contained in one big memory. No setting up the machine. You walked up to it, you fed your program in, and then
you were in business. You didn't have to set a lot of switches, or knobs, or things of that sort. In fact it was the idea of the modern computer.' In contrast to the secrecy surrounding Colossus there was a 'very free exchange of
information'. Wilkes recalled: 'I came back from that course feeling that I knew everything there was to know. Some of the people on the course, and some others, set about building computers.
The Edsac was one of these [projects].
A major technical problem was how to implement fast memory. At Manchester University, FC Williams and Tom Kilburn solved the problem by using a cathode ray tube. Their 'Baby' was working
before Edsac but it was a short- lived pilot project.
Wilkes had to build a computer that could be used in earnest. At Eckert's suggestion he used magnetic delay lines as memory. Building a computer was a
'very large-scale engineering problem' requiring unprecedented build and component quality. 'There was a period of some months when all the hardware was there and we were making it work. We had to make modifications where we had not matched up to these high
standards. You see in a computing machine the numbers change very rapidly and when a change takes place you mustn't lose any pulses and you can't have any intrusive pulses.'
How did he feel when it finally worked? 'We all made a beeline for the pub and celebrated. Success for Wilkes meant
I moving to a second phase. 'The Edsac project was two things. It was to build a computer that was workmanlike - not
necessarily the best that might with more time be done with the technology of the day - so that we could get on with I running some problems and getting experience of its use...we began to turn our attention
to the development of what is now calling programming I methodology.
Wilkes is self-effacing about the fact that his team beat the Americans in building the first stored-program computer to go into service. There was not a race, he said. Or if there was, it was one in which the competitors each had a different winning post.
'Eckert and Mauchly, the Eniac engineers, set up a company to produce a computer they could sell for business purposes.
Clearly, they had to achieve a higher degree of technical perfection than ours. We cut a lot
of corners to get things going. There was a sense among all people working in the field at the time that computing would have very, very wide applications'.
With early machines breaking down several times a day, what was not foreseen was that computers would get reliable, Wilkes said. Valves 'weren't particularly
unreliable but resistors tended to drift in value and soldered joints would
give a lot of trouble.
'Of all the advantages of integrated circuits -low cost, high speed and all that - the most important thing it seems
to me is reliability.
The idea that you can have a computer of your own and that it will work for months or years without going wrong. The press got hold of the Edsac story and had a field day,
with much talk of a 'mechanical brain' - which was curious, considering that mechanical computers were precisely what
electronic ones were superseding.
Also curious was the fact that Alan Turing, the only UK computer pioneer of the time to be famous today, was unknown outside academic circles. Wilkes, who was the same age as Turing and went up to Cambridge with him in the same year, recalled: 'I liked him. He was a quiet man who kept himself to himself. But he said Turing's post-war
work
at the National Physical Laboratory, and later at Manchester University, met
little success in practical terms. 'He wasn't a man of action. He didn't know
how to get a project going. I don't think The old way. I don't think he was very
interested in the user side of a computer. He was more interested in ideas. His record of achievement is really
he very slight.
Wilkes headed the Computer Lab until 1980, and the spent five and a half years with Digital Equipment in Massachusetts
before returning to Cambridge. The extraordinary thing is that progress was enormously fast in those first 10 to 15 years and in the last 15 years it has I been as fast or even faster. People were always saying that the computer field will settle down and it will become like the motor car industry or something. But it hasn't. It has it retained its excitement.
Sir Maurice Wilkes' 1985 book Memoirs of a Computer Pioneer (MIT Press ISBN 0-262231-22-0)
is still in print and costs £23.50 from John Wiley distribution (01243 779 777).
Wilkes and colleagues in the 1930s with a mechanical analogue computer that was designed to solve differential equations. Analogue computers can also be electrical and may still be used for some purposes
top>>