The Atlas Laboratory in the 1960s was essentially a supercomputer centre, although the term supercomputing had not yet come into general use. The Ferranti Atlas was a supercomputer of its time: a machine that had processing power greatly above the norm but nevertheless was designed to be able to support a broad range of demanding applications rather than being targeted at any one specific discipline. For several years the Atlas met the heavy computational requirements of a wide range of scientific projects, almost all from disciplines other than particle physics.
The very powerful IBM 360/195s which were housed at the Atlas Centre between 1971 and 1982 were provided primarily to meet the requirements of particle physics although twenty percent of their capacity was available for other disciplines funded by the SERC. To some extent this enabled the 'Atlas users' to continue with their computational work and further opportunities opened up for them with the provision of a service on a Cray-1 supercomputer at the Daresbury Laboratory which ran from 1980 until it was transferred to the University of London Computer Centre in 1982.
Dedicated supercomputing returned to the Atlas Centre in 1987 when a Cray X-MP/48 was installed to meet the needs of the most computationally intensive projects from the academic communities supported by all the UK Research Councils. Shortly afterwards IBM invited RAL to be the UK participant in its European Supercomputing Initiative. In consequence the Laboratory's IBM 3090 mainframe was expanded by adding processors, vector processing capabilities and extra memory. The peak performances of the Cray and IBM machines were 940 and 696 MFlops respectively, the IBM machine having the larger memory. With these two facilities the Atlas Centre had, for a time, the most powerful computing equipment for academic science in Europe.
In November 1992, after completion of the IBM Agreement, most of the supercomputing attributes of the IBM 3090 were removed. However, in October that year the Cray supercomputing service had expanded when the X-MP was replaced by a Cray Y-MP which had about three times the computational power, and this machine would continue to serve the Research Councils' computational scientists until 1996.
