A Potted History of early IBM Computers at Daresbury

IBM 1800

For general information about the IBM 1800 Data Acquisition and Control System see Wikipedia and IBM history site.

The 1966 comparison by Davies and Zacharov [DZ1966] describes a proposed upgrade to the system as follows.

The central processors and adaptors of this computer are manufactured in Scotland.

The 1800 is designed specifically for data acquisition. The memory word length is 16 bits plus a parity bit and a memory protect bit, the memory cycle time is optionally 2 or 4 us. Instructions may be of one or two words, there are 32 basic operation codes and use of modifiers gives over 100 instructions. Hardware instructions allow for indexing, displacement addressing and direct or indirect addressing. The following features are standard: 3 data channels operating on a cycle stealing basis; 3 index registers; and 12 interrupt levels each allowing 16 separate interrupts. Hardware multiply and divide is standard.

Low speed input/ output takes place by execution of program instructions. At high speeds transfer is initiated by an instruction and proceeds by cycle stealing. One cycle is required per word; in the 2us option the maximum rate is 500 Kc/s.

The software includes an assembler language, Fortran which is a subset of the 360/50 Fortran, and a subroutine library. A time sharing executive is available for execution of non-process programs during a data acquisition program, but at least 19K of core store and a disk core required.

The quoted cost of configuration I [as given in the introduction to the report] is £34,049 for 4us memory cycle including the IBM 20% academic contribution, and £30,192 exclusive of import duty. The magnetic tape unit is 9 track, 37.5 i.p.s., 80 b.p.i. single drive. The quotation includes provision for up to 48 interrupt points and a 360/50 adaptor. IBM recommends the inclusion of a 1442 card read/ punch. This is because configuration I is too small to use the time sharing executive, and the card based system used is slow using the paper tape input provided. The cost of the 1442 is an additional £4,892, including import duty. The 2us central processor is £2,348 more expensive than the 4us option.

Servicing and maintenance facilities already exist at Daresbury for the existing 1800. The delivery time is about 18 months.

Control and monitoring of the Nina synchrotron was carried out using the IBM 1800 installed in the central control room. This system recorded nearly 200 machine parameters every minute, such as vacuum pressures, magnet currents and r.f. power levels. These values, which could be printed out on demand, were stored in the computer for periods up to 4h, so that a continuous record of machine behaviour is available for fault diagnosis. An alarm facility could be actuated if any of these values went outside preset limits, and it was hoped that the computer would be used to control the values of some of the more critical parameters directly.

IBM 360/65

The System 360 was a 32-bit machine introduced in 1964. Both the 1800 and 360 were programmed in FORTRAN. For more information see Wikipedia. The system was originally an IBM S360/50 and was updated to a 360/65 in Nov'1968.

Scientists from the universities of Glasgow, Liverpool, Lancaster, Manchester and Sheffield worked with resident staff to run experiments on the Nina synchrotron. Data was analysed on the IBM S360.

The following photo dates from 1969.

IBM 370/165

Some more pictures and facts for the IBM 370/165 at Daresbury in 1975. The System 370 introduced by IBM in 1970 was also a 32-bit machine, see Wikipedia.

The following press release is from Computer Weekly 6/7/1972. [note some transcription errors.]

All the signs are that Daresbury Nuclear Physics Laboratory, near Warrington, Lancs, is to be considerably upgraded as a scientific research centre.

Government approval has already been received for the installation of an IBM 370/165 to replace the existing 360/65, the Science Research Council have recommended siting a 20MV heavy ion electrostatic generator at Daresbury to be used by the Universities of Manchester, Liverpool and Birmingham and more tentative plans are afoot to develop a 0GeV electron accelerator as well.

The experimental work now being carried out at the laboratory centres around the 5GeV NINA electron synchrotron which enables scientists to carry out some important work on the electroproduction and photoproduction of strongly interacting particles in the energy range 1 GeV-5GeV.

Apart from the staff of the laboratory itself, teams of scientists from the universities of Manchester, Lancaster, Glasgow and Sheffield and from Pisa in Italy are carrying out experimental work on the main NINA apparatus.

Although NINA and its German counterpart DESY in Hamburg are carrying out important work in particle physics, the most glamorous experiments are inevitably carried out on the giant 2OGeV. linear machine at Stanford University, California, where an electron beam is accelerated along a tunnel several miles long before smashing into a target to probe the inner structure of the fundamental nuclear particles.

The computer installation at Daresbury centres around an IBM 360/65 with 1 Megabyte of fast core store and another half Megabyte of direct access store. An IBM 1802 concentrator handles input from on-line experiments and multiplexed terminals while two 2250 interactive graphics units are used in the study of beam optics.

With a large amount of its power dedicated to on-line working, the computer at Daresbury is really the "engine room" of the laboratory and the new 370/165, which is said to be about three times more powerful than the current machine, will have most of its extra power taken up by increased rates of on-line data acquisition from existing experiments.

Operations at Daresbury have been somewhat restricted because the Science Research Council have been heavily committed to supporting the development of the 300GeV proton synchrotron at CERN, the European centre for fundamental research in Geneva, explains the laboratory's director, Professor A. Ashmore.

However extensions to Daresbury's experimental programme are now being planned and include investigations in atomic, molecular and solid-state structure using the strong electromagnetic radiation which is created by the electrons orbiting in the main accelerator.

Installation of equipment for handling this radiation is nearing completion and university scientists from Manchester, Reading Oxford and Cambridge as well as staff from the National Physical Laboratory are preparing experiments which are expected to go into operation next year.

As well as providing computer services to the scientists from Daresbury and the various universities involved in the experimental work, staff of the laboratory's computing and electronics group also carry out systems development work, particularly in the area of interactive graphIcs and their application to the study of beam optics.

Another focal point of the software development work is the provision of Camac type interfaces for data handling with various small peripheral computers. Camac controllers designed at Daresbury are in use at CERN and Liverpool and Manchester universities, the Lawrence Radiation Laboratory in California and Canada's Alberta University.

In order to improve the job handling characteristics of the main computer, its OS software has been modified so that programs which might adversely affect the processing efficiency if they are loaded into the wrong areas of core are guided into the CPU under operator control.

As with most scientific research establishments, Daresbury frequently has had a large backlog of work-in-hand, but this situation will hopefully be alleviated with the installation of the large 370/165.

In January 1973 the IBM 370/165 therefore replaced the IBM 360/65 which has been in service from 1968. This had a faster processor plus new disk and tape drives (self loading) capable of holding more data. Error correction was also a feature of the new system making it more reliable than previously. In total the on-line storage grew from 500MB to 850MB. This grew to 3.5GB by the end of 1980 and had 3MB main memory.

To enable users to fully benefit from the new system, a new card punch and reader was provided in an area of the computer hall, separated from the main operations area by a reception and output desk. There were also user terminals driven by TSO which could be used for program development, debugging and job submission.

IBM 370/165 console.

Work carried out on the S370/165 was quite varied. It included protein crystallography analysis for CCP4, analysis of NMR spectra [loomes] and marine forecasting of sea tidal surges [nihoul] in association with Bidston Observatory.

to main site.