The University of London ordered its ATLAS computing system from Ferranti Limited to replace their overworked Mercury machine in August 1961. ATLAS was, at that time, the largest and most powerful system on offer in Great Britain. The initial order consisted of :
The total value of the system ordered was of the order of three quarters of a million pounds. The total cost was met from four sources: the University Grants Committee who gave a grant of £50,000 towards the cost, the British Petroleum Company who made a substantial contribution to the capital cost in exchange for about one quarter of the available machine time, the University Funds and from a loan made to the University.
At this time the University set up a company known as Computers (Bloomsbury) Ltd. (now known as London University Computing Services), and it was the intention that this venture would provide a general computing service on ATLAS for commerce and industry in order to try to recover the debt incurred by the purchase of the machine and to contribute to the high cost of running such a machine. A body known as the Atlas Computing Services was also established to be responsible for the operation of ATLAS and its ancillary equipment. The University of London Computer Unit (which became the Institute of Computer Science in March 1964) which had operated the University's Mercury computer was to become responsible for research into computer science, teaching, giving advice on programming techniques and channelling the University jobs to ATLAS. So the order was placed, the administrative structure prepared and a period of waiting with trepidation for and in anticipation of the future began.
The Manchester ATLAS computer came into use (without drums) in January 1963 and gradually a service developed on the machine there. A courier service was operated from London to the Manchester computer to frequent programmers in London with their new servant (master?). During August 1963, the London ATLAS was seen in action at the West Gorton factory with a program that used four paper tape readers, two paper tape punches and two line printers at the same time. The official switch off date at the factory was to be the end of September, but this was delayed at the recommendation of Ferranti to allow further commissioning to be undertaken.
The building to house the London ATLAS had been completed in July 1963 and men from Ferranti were engaged in laying wiring ducts and cables since then. This building was a two-storey unit, the top floor was to house the computer's peripherals and the basement, the central computer, stores, drums, co-ordinators, engineer's console and power regulators. The reader must remember that because ATLAS was built using discrete components the machine itself was very large and demanded high power levels. Of special interest must be the spiral staircase that connected the two levels of this building - this surely must have been the only computer room to boast its own spiral staircase.
Before ATLAS was delivered to London a transformation took place: the computer interests of Ferranti Limited became part of International Computers and Tabulators Limited (which is now International Computers Limited). This involved a great deal of extra work to be carried out on ATLAS. The names on all the cabinets had to be changed from Ferranti to I.C.T..
At the end of October 1963 the delivery of the London ATLAS began. A crane was moved into position and the motor alternators (which provided the power at a more suitable frequency and voltage for various parts of the machine) weighing between three and four tons were lowered into their new homes under the supervision of I.C.T. engineers. The computer itself arrived in the middle of November together with the consoles, fixed store, working store, some peripherals and the peripheral co-ordinator.
A week later, more of the machine arrived including the core store, store co-ordinator, line printers and magnetic tape decks. Various delays seemed to have occurred especially with delivery of the four magnetic drums which were essential to the system.
By January 1964, the commissioning of the computer on the site had begun; this was to be a long and somewhat tedious process for all concerned. An extension to the original order had been made in 1965 to include a further 16K of core store and a new store co-ordinator and these were to be effected at the end of 1964.
On the 11th May 1964, the computer was handed over for a shift per day.
During the first week of this regular access to the machine, it was promptly handed back to the I.C.T. engineers as it did not appear reliable enough for any serious work to be carried out on it. Gradually this situation did improve as the hardware and software shortcomings were corrected and small quantities of work was processed.
On June 4th 1964, Her Majesty Queen Elizabeth, the Queen Mother, visited the Institute of Computer Science and the Atlas Computing Service in her official capacity as Chancellor of the University. During this visit Her Majesty was shown around the ATLAS installation and two demonstration programs were run on the ATLAS computer. One program was a concordance program for St. Mark's Gospel, and the second a program that printed anagrams of Clarence House, culminating in a line- drawing of Clarence House itself.
By the end of September 1964 the service provided on the machine was becoming more acceptable. The lack of fully developed software, however, was causing some concern. In October, the computer was shut down to allow the extensions to the core store to be made. (During this closed period some University work was carried out on the Manchester ATLAS.) Then the London ATLAS was back on the air again, most of the available time was given over to testing the machine's performance. At this time B.P.'s test program was getting runs of between four and five hours, but the magnetic tape system appeared to be falling short of its expected performance. By February 1965, B.P. were running 72-hour test sessions with only occasional trouble from core store parities - a problem that was to remain with the machine throughout its life.
In March of 1965, the system seemed to be coming out of the wood and developing into a viable service.
During April 1965, two thousand jobs were run on the machine in one week and gradually this figure rose to around three thousand eight hundred jobs per week in March 1966.
Further enhancements were made to the system in the form of two IBM compatible magnetic tape decks. These were brought into service around April 1966, so allowing ATLAS to read and write IBM magnetic tapes.
During its long life, the London ATLAS ran many thousands of jobs, not, however, without problems. Besides a certain degree of difficulty over administration of machine time, two main problems emerged out of the hardware. The most serious of these was the problem with the 1in magnetic tape system which never achieved its expected performance, and since this part of the hardware was of a very high degree of importance to the system as a whole, the performance of ATLAS suffered. There
were also problems with the core store units, though these were far less serious (and less frequent) than the magnetic tape problems.
The London ATLAS was switched off for the last time on the 30th of September 1972 and scrapped. Parts of the computer were, however, being sent to museums as exhibits for technological sections.
The University of London replaced the ATLAS with two CDC machines, a CDC 6600 and a CDC 6400, and these were gradually brought into service around 1970-1971.
London University Computing Services were allowed to buy a CDC 6500 computer and still continue to provide an extremely high quality commercial computing service.
