This Atlas site has variously been referred to as NIRNS, Harwell, Chilton or RAL, so a word of explanation is helpful. The National Institute for Research in Nuclear Science (NIRNS) was formed in 1957 to operate the Rutherford High Energy Laboratory as an open-access nuclear research facility, located just outside the perimeter fence of the UKAEA's restricted-access laboratory at Harwell. NIRNS itself lay between the villages of Chilton and Harwell, so 'Chilton' became the informal name for NIRNS.
In the summer of 1961 the UKAEA placed an order with Ferranti for an Atlas computer, costing about £3.1m. Chilton was chosen as the site, so that the computing facilities could be made available to outside users - and in particular to the UK's academic community. The original intention had been for the Chilton Atlas to be used in equal proportions by Harwell, Rutherford and the Universities [ref. 5]. However, by 1964 both Rutherford and Harwell had acquired their own IBM facilities so the Chilton Atlas was mainly used by the universities and other government departments, such as the Meteorological Office and the Natural Environment Research Council (NERC).
At Chilton, a special Atlas Computer Laboratory was designed and built to meet the needs of a large computing service. In 1961 Jack Howlett was appointed Director of the Atlas Laboratory. The new building was ready for occupation in January 1964 and the machine was installed during May and June of that year. In October 1964 a regular 'at risk' service was started, consisting of one 8-hour shift per day. (The remainder of the time was spent on software development and engineering maintenance). In May 1965 the formal hand-over and acceptance period began, leading to full three-shift operation and final acceptance by May 1966.
The Chilton Atlas was closed down on 30th March 1973 and the main hardware units were presented to the National Museums Scotland, where they remain in storage. The original plan was to replace Atlas by a twin ICL 1908A computer, which was estimated to be 20 times the power of Atlas. In the event, ICL cancelled the 1908A project and Chilton acquired the less-powerful 1906A computer plus a 20% share of the adjacent Rutherford Lab's IBM 360/195. The replacement facilities for the Atlas at Chilton started to be assembled in July/August 1971, with the delivery of the ICL 1906A. This was housed in an adjacent new computer block, completed in the summer of 1971. At that stage there were about a hundred staff. The Chilton 1906A had a main core store of 256K 24-bit words. Computing staff at Chilton rated the 1906A at between 2 and 3 times the speed of Atlas. The 1906A and Atlas ran a parallel computing service between October 1971 and March 1973.
Through a series of mergers in the period 1975 - 1979, the Atlas Computer Laboratory became part of the Rutherford Appleton Laboratory (RAL). In 2007 the Science and Technology Facilities Council (STFC) took responsibility for RAL. Today, RAL has a staff of about 1,200 people who support the work of over 10,000 scientists and engineers, chiefly from the UK's university research community. At Chilton the computing facilities in the recent past have included two IBM 360/195s, a Cray X-MP and a Cray Y-MP. STFC's computing facilities are frequently upgraded. At the time of writing, STFC's largest facility (and the UK's most powerful supercomputer, see ref. [1]) is an IBM BlueGene system (named Blue Joule) at Daresbury.
The following technical details of the Chilton Atlas are largely based on the collection of documents assembled at: http://www.chilton-computing.org.uk/acl/ with additional help from Bob Hopgood who worked at Harwell and Chilton from 1959 - 2000.
The Chilton Atlas was the largest of the three Atlas 1 installations. In 1966 it consisted of the following facilities [ref. 2]:
Primary memory (all capacities given in terms of 48-bit words): 48K core at 2 µsec. cycle time; 96K drum; 8K fixed store at 0.4µsec. access-time; 1K Supervisor working store at 2 µsec. Magnetic tapes and other peripheral equipment: Sixteen Ampex TM2 decks (1-inch); two IBM 279 Mark IV decks (half-inch); two ICT card readers (600 cards/min.); two Ferranti paper tape readers (300 chars/min.); two Anelex lineprinters (1,000 lines/min. 120 chars/line); one card punch (100 cards/min.); two Teletype paper tape punches (110 chars/sec.).
The installation occupied a total area of 12,100 sq. ft. on two floors, with additional space allocated for programmers' offices, etc. On the upper floor the Atlas input/output equipment and magnetic tape decks occupied one large room, with the data-preparation equipment in an adjacent large room. The Atlas processor and memory system (including drums and, in due course, a large file disk) occupied one large room on the ground floor, with the motor-alternator and the air conditioning plant in an adjacent large room.
The available high-level languages included Fortran, Algol, LISP, IPL-V, Atlas Autocode and Extended Mercury Autocode. In 1966 approximately 65% of programs (by number and by computing time used) were written in Fortran. In 1966 the applications areas, as a percentage of the total number of jobs run, included Engineering (31%), Physical Sciences (29%), Mathematics (23%), Biology and medicine (5%) and Social Sciences (5%).
By 1966 the Atlas Computer Laboratory was processing an impressive number of programs each day. The Director, Jack Howlett, was able to write in his annual Report: "In a typical week we run 2,500 jobs, input 800,000 cards and 30 miles of paper tape, print 1.8 million lines of output, punch 50,000 cards, handle 1,200 reels of magnetic tape. We have 250 projects on our books from university users and are usually doing work on 100 of these. Our experience over the past year has shown that the Atlas central processor, with the Supervisor which is an integral part of the system, is an exceedingly powerful and flexible device which deals smoothly and efficiently with a heavy load of very varied work". Mike Bayliss, who went on to develop the Chilton time-sharing system with John Baldwin and Eric Thomas (see below), commented in 1966: "In some respects Atlas is ideal; there is a sophisticated store management system, a large virtual memory, good interrupt facilities; Atlas is a very fast and reliable computer". [Ref. 3].
However, in the light of hindsight it is now apparent that programmers' expectations were gradually changing. Encouraged by experiments in multi-access timesharing facilities being conducted at MIT on an IBM 7090 computer [ref. 4], the emphasis was slowly moving away from efficient use of mainframe hardware and towards efficient use of programmer's time. By 1966 the Atlas Computer Laboratory was considering how to improve the human-machine interface.
From 1966 to 1970 the Chilton Atlas was improved in two areas: (a) the addition of specialist peripheral devices; (b) the provision of better file-manipulation facilities and time-sharing via a number of interactive terminals. The latter enhancements required investment in two major pieces of hardware: a large disk store and a satellite computer. The Atlas Supervisor proved robust and flexible enough to take these enhancements in its stride. The changes to the Supervisor required to accommodate the disk were made by ICL and the software for the multi-access system was designed and produced at Chilton.
The large disk store, added in February 1968, was a Data Products model 5045, of capacity 16.8 million words with two independent read-write mechanisms. These gave dual access from both Atlas and the Sigma 2 satellite computer (see below). The transfer rate for blocks of 512 48-bit words was 6.5 milliseconds, with an average access time of 195 msec. In addition to supporting timesharing via the satellite, the Data Products disk improved the performance of the main Atlas computer by taking the place of the four Ampex magnetic tape decks normally used by the Supervisor for the input/output wells and for system software such as compilers and library programs.
The satellite computer, added in 1968 and operational in support of timesharing by 1969, was an SDS Sigma-2 (marketed in the UK by GEC under the name S-2). This machine had a core store of capacity 32K of l6-bit words. The Sigma-2 communicated on the one hand with Atlas through the disk and also through an independent channel, and on the other hand with a number of online terminals. Initially there were six ASR 35 Teletype consoles, VDUs not being available at that time. Eventually there were 32 online terminals, all of which were Cossor DlDS 402 VDUs. Of the 32 terminals, up to eight could be active, four of which could be remote from the Chilton site.
The Sigma-2 relied on Atlas for the actual compilation and execution of users' programs but performed all the other tasks associated with a timesharing system locally. These tasks included: control of the online terminals; provision of a comprehensive filing system using the large Data Products disk; provision of an interactive command language. The Chilton timesharing system was released in May 1969 after a total development time of approximately 24 man-months of work [ref. 5]. In 1971 a fixed-head disk was added to the Sigma-2, to help with the accumulation of output.
Turning to specialist peripherals, a Stromberg Carlson SC4020 Microfilm Recorder was purchased in 1967 at a cost of about £70,000. This produced text at speed and could draw 10,000 vectors a minute. A special attraction was its ability to make animated films. The SC4020 generated output based on a set of commands that it received on an IBM 7-track magnetic tape. One of the major uses of the two IBM decks on the Chilton Atlas was the generation of output tapes destined for the SC4020. (Another major use was in the conversion of satellite-tracking data from IBM tapes to Ampex tapes and then back again). The SC4020's original 16mm camera was replaced by a pin-registered Vought camera with both 16mm and 35mm movements in the Autumn of 1971, with new lenses installed in 1972.
In February 1969 a case was made for purchasing an interactive graphics system. It was planned that the major application would be previewing SC4020 output. Having a small system with graphics display and magnetic tape deck would make it possible to check SC4020 output before wasting valuable SC4020 time and processing capacity. The two front runners for Chilton's interactive graphics equipment were an Elliott Automation 503 system and a PDP15 system from the American company Digital Equipment Corporation (DEC). The Elliott system proved to be more expensive and its display had a relatively slow orange phosphor, so the PDP15 system was eventually acquired - (something of an administrative triumph for Chilton, in an age when it was government policy to 'Buy British'). A basic PDP15 computer with a VT15 display was purchased in 1970 and gradually enhanced with a disk, additional memory, a high speed interface to the Chilton 1906A and a connection to a VCS3 synthesiser.
Other special devices installed at Chilton over the period 1966 - 70 included a D-MAC graphical input table, an Opscan Optical Character Recognition device and an Optronics Microdensitometer (operational in about 1972).
At the Chilton Atlas close-down ceremony on 30th March 1973 the Director, Jack Howlett, was able to report as follows. Of the 44,500 scheduled hours of computing time during the period May 1965 to March 1973, 43,000 hours had actually been usable - yielding an Atlas availability of 97%. The central processor usage statistics during this period were: User programs: 82% of CPU time; Supervisor activity: 12% of CPU time; Idle time: 6% of CPU time.
From May 1965 to March 1973, 863,000 jobs had been handled by Atlas, with a total market value of computing time of £10.8m. Approximately 85% of available computing time had been devoted to UK universities, during which 2,300 research projects had been supported. The remaining 15% of computing time was used by government departments for applications such as weather forecasting and space research. About 70% of all programs used Fortran, with the average run-time for a job being about 150 seconds of Atlas CPU time.
Over the years, the Chilton staff found that there were certain academic areas where the needs of most users could be met by general programs, which staff then provided - either writing them or acquiring (and adapting) them from other sources. Amongst such area were:
Crystallography: Determination of crystal structures by interpretation of x-ray diffraction patterns. In (** date??) equipment was added (** to Atlas?) that scanned the films and produced the digital information needed as input to the analysis suite.
Computational Chemistry: several program suites were implemented and supported to calculate the structures and properties of molecules and solids.
Finite-element analysis: Mostly for engineering structures.
Time-series analysis: Used to detect patterns or variations in recorded phenomena such as temperatures, river flows, power flows in the national electricity grid system, elecro-encephalograph recordings, etc.
Text analysis: Investigation of quantitative features of natural-language texts, such as word count, vocabulary, frequency distribution of word use, word length and sentence length.
Survey analysis: Operations such as counting, classifying and correlating the information gathered in surveys. Applications include the study of medical or social questions, such as the incidence or spread of some disease in a particular industry or geographical region.
Overall, the Chilton Atlas was judged to have provided a highly-valued service to the UK's scientific base for about nine years. It is impossible to mention all the Chilton staff who contributed to this success but amongst them were: Jack Howlett (the Chilton Director) and under him Bob Churchhouse (programming) and Jim Hailstone (operations). Under them were Barbara Stokoe (user/application support), Paul Bryant and Mike Bayliss (operating systems support), Eric Thomas (Sigma 2), Bob Hopgood (graphics and compilers), Doug House (Atlas operations), John Baldwin (chemistry applications), and a plethora of associated UK academics.
Ref. 1: A periodically-updated list of the world's 500 most powerful supercomputer installations will be found at: http://www.top500.org/
Ref. 2: J Howlett, Progress Report on NIRNS Atlas 1. 18th October 1964. See:
http://www.chilton-computing.org.uk/acl/literature/papers/p007.htm
Ref. 3: M Bayliss, Time sharing and Atlas: Some Aspects of Current Operation and Research. 1966. See:
http://www.chilton-computing.org.uk/acl/technology/sigma2/p001.htm
Ref. 4: F J Corbato, M M Daggett and R C Daley, An Experimental Time-sharing System. Proceedings of the AIEE-IRE Spring Joint Computer Conference, May 1962, pages 335 - 344.
Ref. 5: J Baldwin and E Thomas, Multi-access on the Chilton Atlas. Computer Journal vol. 14, issue 2, 1971, pages 119 - 122.
Ref. 5. Bob Hopgood, e-mail to Simon Lavington, 15th August 2012.
