Contact us Heritage collections Image license terms
HOME ACL Associates Technology Literature Applications Society Software revisited
Further reading □ OverviewAtlas computer technology (David Aspinall, 2001)Atlas Computer Family (Bob Thorley, 2015)Memories of Atlas Fortran (Ian Pyle, 2012)My life on Atlas 1 and 2 (Brian Chapman, 2014)Memories of Bernard Loach
ACD C&A INF CCD CISD Archives Contact us Heritage archives Image license terms

Search

   
ACLLiteraturePersonal
ACLLiteraturePersonal
ACL ACD C&A INF CCD CISD Archives
Further reading

Overview
Atlas computer technology (David Aspinall, 2001)
Atlas Computer Family (Bob Thorley, 2015)
Memories of Atlas Fortran (Ian Pyle, 2012)
My life on Atlas 1 and 2 (Brian Chapman, 2014)
Memories of Bernard Loach

The Atlas Computer Family -- an overview

Robert (Bob) Thorley

2015

Atlas, one of the world's first "supercomputers", was the result of intensive on-going collaboration between Manchester University's Department of Electrical Engineering and Ferranti Computers Ltd. On entering service, the first Atlas became the world's fastest and most powerful. Indeed, it was calculated that whenever Atlas was off-line, for example for maintenance, the UK lost half its computing power.

Predecessors to Atlas (like the Ferranti "Mercury"), had been severely limited by their reliance on power hungry, bulky and unreliable thermionic valves. However in the late 1950s, new technologies enabled the development of ground-breaking "number-crunchers", like Atlas. These breakthroughs were principally the availability of fast, reliable and inexpensive transistor switching circuits (using germanium semiconductors), plus large capacity magnetic core storage. At last, mainframe visionaries could realise the scale, power, architecture and software sophistication, which had previously been mere aspirations.

The chronology of the Atlas family encompassed two distinct generations over 17 years:
Atlas 1 computers were installed at Manchester University (1962), London University (1963) and Chilton (1964). Ferranti then in collaboration with Cambridge University Maths Lab (Computer Science Dept.) undertook the development of Atlas 2 -- a less expensive hardware configuration, with dramatically advanced and sophisticated software.
Atlas 2 computers were installed at Cambridge University (1962), AWRE Aldermaston (1964) and the CAD Centre in Cambridge (1967).

Whilst successfully meeting much of the need for large scale computing power within the UK, regrettably none of the planned export sales were achieved. The last Atlas closed down in 1976, by which time Ferranti Computers had merged with other UK rivals to become the more commercially competitive ICL. Here Atlas technology and expertise lived on in the successful "ICL 1900" range of computers, and particularly in the successor to Atlas, the 1906A mainframe.

The Atlas design teams pioneered many ground-breaking innovations now mainstream in computing. These, including virtual memory, multitasking and time sharing, set standards for others to emulate. The select band of British academics and engineers who created these spectacular advances, were giants indeed.

In an air-conditioned Atlas computer room one would observe a control desk, peripherals like punched card and paper tape equipment, perhaps a dozen, one inch magnetic tape decks, magnetic disc or drum storage, etc. However central to the whole assembly were rows of six foot tall cabinets containing the Atlas processor hardware and core memory. Within a typical refrigerated air cooled cabinet were 10 horizontal racks, each containing 48 circuit boards plugged into 12 pin sockets at the rear. These connected to the enormously dense and colourful back-wiring, inter-connecting each element of the system. There were 40 standard types of logic module/circuit board, each identified by a 3 digit type number visually coded by coloured plastic tags -- using the same numeric colour convention found on resistors. By the use of an oscilloscope, the operation and status of any circuit (logic element) on a board could be conveniently monitored in situ in its rack.

One would observe that these modules could not have been more tightly populated and packed, this to minimise the distances signals had to travel in order to eliminate unnecessary delay. (At this level of processing performance, it really mattered that (at the speed of light); it takes 1 microsecond for a signal to pass along a 1 foot length of wire. (These were pre-metric days of course!)

The engineer's working blueprints defined the operation and relationship of each element of system hardware (for example: within the central processor, floating point unit, memory, storage, peripheral control, etc). Here one would find the purpose and operation of each circuit and its place in the overall system schema. Meanwhile the supremely complex back-wiring dimension was managed through a bespoke Computer Aided Design program.

This snap shot of exciting times, is drawn from the reflections (from a Ferranti perspective), of a young Atlas 2 (and then 1906A), design and commissioning engineer.

Copyright Robert Thorley 2015

⇑ Top of page
© Chilton Computing and UKRI Science and Technology Facilities Council webmaster@chilton-computing.org.uk
Our thanks to UKRI Science and Technology Facilities Council for hosting this site