Computers and computing
The work of CERN is a very small and specialized part of contemporary life but subnuclear physics research could not have sustained its rate of progress had it not been for the parallel development of computers.
The designers of the early computers assumed that programming would be done by small groups of specialists, probably mathematicians, and that it would be undesirable to make the task too easy. For example, von Neumann and Goldstine, who in 1946 proposed what is essentially the modern computer, argued against built-in floating-point arithmetic: "The floating binary point represents an effort to render a thorough mathematical understanding of at least part of the problem unnecessary, and we feel that this is a step in a doubtful direction."
The first successful use of electronic computing to analyse data from a particle-physics experiment was made in 1956 by a group at Berkeley, using a cloud chamber in a beam from the Bevatron.
CERN's first electronic computer began operation in autumn 1958, a Ferranti Mercury; it taught CERN two lessons. Firstly, installing new, large computers brings serious unforeseen difficulties: late delivery and an initial period of recurring hardware and software problems. Secondly, the ready availability of a computer stimulates computer use at an alarming rate; Mercury had barely completed its first year of operation when CERN had to order its second computer, an IBM 709.
With Mercury and the 709 operating together, CERN had its first experience of compatibility problems. This is a continuing source of difficulty as more and more different computers are coming into operation at CERN and many of CERN's programs are used on different computers in Member States.
Fortran made its debut at CERN with the IBM 709, and the use of this language was encouraged so that today it is essentially the only language in general use.
In 1960, the first ideas for flying spot digitizers to measure bubble-chamber film began to take shape. The prototype device was operating on-line to the IBM 709 in 1962, beginning the trend towards on-line use of computers for processing experimental data. It also had the less enviable distinction of being our first experience of another hard fact of life, namely that adequate software is as essential as hardware – and much more difficult to achieve.
Before being given to the Institute of Mining and Metallurgy at Krakow University in 1966, the Mercury was connected, via a one-kilometre data-link, to counter and spark-chamber equipment in the synchrotron experimental hall, becoming the first computer used on-line at CERN to analyse experimental data in "real-time".
About 10 small computers are now being used for data-acquisition in electronics experiments in the synchrotron experimental halls. If small computers continue to become simultaneously faster and cheaper, the use of a big computer may become the exception rather than the rule.
In any one week, about 400 scientists and engineers use the CERN central computers. In the next five years, this figure may increase by 50% or more.
Typically, when writing and testing a program or making short calculations, a user would like to obtain results within a few minutes; when carrying out long calculations with a working program it is usually adequate if results are obtained after a few hours or even days.
The time spent in tape handling will reach alarming proportions over the next few years unless some alternative means of bulk storage can be used for data requiring frequent or random access.
Computing facilities in the 1970s will most probably be based on a network of several computers and data stores, with different degrees of connection between them. The computing capacity will have to be accessible from different parts of the site, with enough redundancy that computing continues when some machines or storage elements are out of action.
To restore man's ego in the face of electronics, it is a pleasure to tell the story of a human computer, William ["Wim"] Klein, a member of CERN's Theory Division since 1958. Wim Klein has been a calculating prodigy since his early school days and has entertained audiences for many years. Multiplication of any five-figure numbers takes a few seconds; even 1,388,978,361 × 5,645,418,496 = 7,841,364,129,733,165,056 he did completely in his head in 64 seconds.
• From the articles on pp162–175.
Compiler's Note
The September 1967 issue of the CERN Courier contains 26 fact-packed pages of valuable source material for IT historians. It charts the pioneering approach taken to a broad range of emergent computer applications, from automated film measurement to administrative data processing. The computing culture that was rapidly taking shape throughout the particle-physics community at that time would lead, barely 20 years later, to the creation of the World Wide Web at CERN. One might say that contemporary life today could not sustain its rate of progress had it not been for the development of subnuclear physics research.