Queens University, Belfast, have budgeted £250,000 to expand their computing facilities.
The proposed machine will be used to extend a current programme of theoretical science research and investigations in computing science.
Computing facilities at the university include a Deuce, purchased on a grant from the US Navy, and a data communications link to NIRNS Atlas 1.
As part of the computer expansion Queens are to establish a Chair of Computing Science. It is understood that a computer scientist returning from the US is to be appointed.
A standing advisory panel on computers has been set up by the University Grants Committee in connection with the Department of Education and Science and the Scottish Education Department, it is stated in the committee's annual survey.
The panel will advise the committee and the departments on the proposals for the provision of computer facilities in institutions of higher education. Its members are: Sir Willis Jackson, chairman, Professor of Electrical Engineering at the Imperial College of Science and Technology, London; Dr Gordon Black, a member of the United Kingdom Atomic Energy Authority; Dr E. T. Goodwin, of the mathematical division of the National Physical Laboratory; and Dr J. Howlett of the National Institute for Research in Nuclear Science.
Dr Christopher Wilson, the scientist in charge of the building of the Atlas computers, hit back last night at attacks on the British equipment installed in universities.
He had just watched the criticism of the operating efficiency of British-built computers, including the giant Atlas, on the BBC-2 television programme, Horizon.
In the programme, university men alleged that the computers, Atlas and English Electric models were inefficient, broke down frequently and did not compare with American rivals.
Dr Wilson denied charges that the London University Computer Unit's Atlas breaks down on average every quarter of an hour.
Operating the equipment called for high experience and discipline, and the university authorities were still learning. But the London University Atlas was giving 95% efficiency in 16 hours operation a day.
Both the British Petroleum Co., which has a stake in the London University Atlas, and the National Institute for Research in Nuclear Science at Harwell reported yesterday that they had no complaints with their computers, said Dr Wilson.
The British idea of the efficiency of American computers was distorted in this country, he added. The Atlas was still among the most powerful in the world and highly competitive with American rivals.
The BBC went ahead with its programme, despite last-minute Government warnings of the effect on the morale of the industry.
The Atlas Computer has officially been taken over from the manufacturers, ICT Limited. This decision was reached at a meeting on 15th July 1965 when, after reviewing the performance of the machine over the past three months, it was agreed that takeover should date from 19th May. This does not complete the contractual negotiations; the succeeding year is a period of probation and only if at the end of this we are satisfied with the machine's performance do we finally accept it and make the last payment.
When I wrote about Atlas in last November's ORBIT I said that we hoped to take it over before the end of 1964, adding that it was rash to make predictions in public. It did prove rash. Progress had been very good in the latter part of 1964, but towards the end of the year it slowed down and in early 1965 even seemed to have come to a stop. The trouble was, basically, the interaction between hardware and software. The machine itself, the hardware, is a large and exceedingly complex assembly of electronic circuits and electro-mechanical devices such as card readers and punches, magnetic drums and magnetic tape decks. Its operation as an integrated system is controlled and monitored by an equally large and complex program called the Supervisor. The two interacted in a most complicated way, so that it was often impossible to say whether a fault was due to a failure in the hardware or to a mistake in the Supervisor program; and as the simpler sources of trouble were cleared out, diagnosing and correcting those that remained became increasingly difficult.
Everyone concerned had a pretty anxious time during the first three months of 1965 and when the contract date for take-over, 31st March, approached I had the distressing task of saying that I would not accept the machine: it was not reliable enough. Even so, it was getting through a large amount of useful work every day and we had got on well with the development of Fortran and Algol compilers and with the production of library programs. In April things began to change rapidly and performance to improve dramatically. This was clearly a threshold effect; once the number of faults had been brought below some critical level the engineers and systems programmers could see what was happening and tackle the faults which then arose with much greater understanding and chance of success. We set a target for standard of performance which we considered would justify acceptance and at midnight on 19th May this was attained.
A few figures will give an idea of the way the Laboratory is now working. We have the machine from 9.30 am to midnight, Monday to Friday; the maintenance engineers take it for an hour at some convenient time, usually around 6pm. We are getting about 90% good time each week, that is about 60 hours out of a possible 67½. The best week we have had so far (5th to 9th July) gave 95.7%. In a typical week we will run 1500 jobs, read 350,000 cards into the machine and print out 1,000,000 lines. Sixty per cent of jobs take less than one minute, 3% more than 30 minutes. The machine operates at about 300,000 instructions per second. About 40% of the time is being taken by the universities, 25% by the Rutherford Laboratory, 15% by the Atlas Laboratory itself, 15% by Harwell and 5% by Government laboratories. Work is being done for 25-30 universities who between them have requested time for just under 350 different projects. The forecast of demand is such that we shall need to schedule the machine for 100 hours per week from October 1965 and for 140 hours (the maximum possible) from some time between April and June of 1966. This pressure puts a premium on efficiency - in operation of the machine, in the Supervisor and compiler programs, and in the working programs written by the users.
International Computers and Tabulators is building yet another of the Atlas computers which have recently been the subject of adverse comment, on grounds mainly of complexity and the time required to bring them into full operation.
The machine valued at just under £1m., is being built at West Gorton and the company says will be held for stock.
This is the sixth in the series, the prototype of which is installed at Manchester University. Half its time is available to the University and half on a service bureau basis to I.C.T. customers in the area.
The first production machine at London University is operating on a two-shift per day basis. During the past four weeks, its availability has fluctuated between 60 and 95 per cent of operating time.
The third and biggest Atlas installed at a cost of over £2m. to tackle the complex computing problems of the National Institute for Research in Nuclear Science is on the way to full commissioning.
It is running an eight to ten hours per day shift and according to Dr Howlett, Director of the N.I.R.N.S. Atlas Laboratories, it is tackling a great diversity of problems including regular operations on theoretical physics and other scientific work for N.I.R.N.S., Harwell and the universities.
The technically advanced Atlas 2, built at Cambridge University, is approaching a mean time between incident rate of three hours on a two-shift basis. For the time being it is largely being used in programme development work by the company and by the Mathematical Laboratory in the preparation of new computer techniques.
The Aldermaston Atlas 2, a £1.5m. complex, has been installed and the finishing touches are being put to operating procedures. This replaces the I.B.M. Stretch computer, in operation at the Weapons Research establishment for several years.
The next step in development of these very large machines will depend to some degree on what the Ministry of technology's Advisory unit may conclude. It is, however, likely that it will merge with development of the largest machines in I.C.T.'s new 1900 range.
The Science Research Council's central computer laboratory at Chilton, Berkshire, has taken over the £2.8m. Atlas computer after which it is named.
This huge machine - the biggest ever built in Britain - which is not due for final acceptance until May, 1966, has gone over to three-shift working and by the acceptance date will be operating around the clock on a seven-day week.
The Atlas Computer Laboratory, based on the one large computer, is one of the most powerful in Europe and is providing a comprehensive service to a large number of research organisations.
These include the Council's own laboratories, such as the Rutherford Laboratory for nuclear physics, which operates the giant Nimrod atom smasher and the Space Research Project. The Harwell atomic energy research establishment, the Meteorological Office, the Road Research Laboratory, the Medical Research Council and research workers in many fields in the universities are also users.
The Atlas at Chilton has an enormous work capacity and to solve the problem of keeping it fully employed an unusual operational system has had to be designed to speed input and output of work.
Input data and operating instructions read from punched cards or paper tape are transcribed on to magnetic tape where the jobs are held in a queue on the input tape.
This feeds work into the system, which is able to perform several operations at once. As results are produced they are directly recorded on magnetic tape. Then they are transcribed on paper tape or punched cards as required. This goes on independently of what other work is being done.
