An intimate dialogue between a computer and a researcher or designer can effectively increase man's creativity and extend his memory. This kind of dialogue is called "interactive computing', in which the designer himself becomes part of the computation - taking decisions and guiding the step-by-step progress towards solving the problem.
Modern Technology is becoming increasingly complex and demands extensive facilities for modelling and simulation. The use of high-speed powerful computers is essential but the factors to be optimised are so involved that human interaction is still needed en route. The Science Research Council has taken a major initiative in this field by setting up a national Interactive Computing Facility which is managed centrally yet consists mainly of computers distributed among the research community.
The scheme involves a pooling of resources and expertise to provide hardware systems, networking and graphics software, and also applications programs. This giant facility is being developed throughout the UK and eventually will be used by over 2000 research workers in engineering and pure science.
In 1975 a Technical Group Report was prepared under the Chairmanship of Professor H H Rosenbrock (UMIST) entitled "Engineering Computing Requirements". The report pointed out that facilities for interactive computing available to engineering faculties in universities and polytechnics were inadequate and proposed that the SRC should step in to remedy this. The Engineering Board of SRC accepted the recommendations of the Rosenbrock Report and gave Rutherford Laboratory the responsibility to co-ordinate and manage the provision of a widely distributed national interactive computing service. Council extended the scheme to make the facilities available to all SRC-funded projects.
To create an Interactive Computing Facility one must establish as speedily as possible a network of computers and communications equipment spread over a wide area to reach the individual users. An important goal is to standardise on the most suitable hardware, and generate all 'systems' software centrally, so avoiding duplication of programming efforts at the different centres. Similarly, where agreement can be reached among research workers in a certain field on the specification of generally useful 'applications' software, a unified programme to work to provide this is more economical than for each university group to try to produce independently their own software tools for complex research projects. Further economy may be achieved by mounting and maintaining specialised sets of user programs at a single 'host' computer only, where the local expertise is greatest in a certain technique. Many jobs must be set up using interactive graphics, then passed to a large batch-processor for heavy calculations and then returned for interactive examination of the results. Extremely reliable communications and systems programs are required to support this kind of networking.
There has been substantial progress in setting up the Interactive Computing Facility which currently serves over 600 users. A main trunk network has been established between Edinburgh, Manchester (UMIST) and Chilton, and there are secondary links radiating from these to reach individual terminals or clusters of terminals at many other sites. From many types of multi-user minicomputers (known affectionately as 'Mums') two have been investigated in-depth and are now defined as the standard for the ICF which will be based upon the use of GEC 4070 and PRIME 400 series computers. It is planned to link about fifteen MUMS into the network some probably with array processors attached and each providing computing 'on demand' for 6-10 users simultaneously. The first new MUM was installed last week at Bristol.
Progress in network techniques is extremely rapid and within 2-3 years, the present core network will be modernised to link over 20 ICF sites using the new International protocol "X25".
Meanwhile every effort is being made to reach more users with the present network. Within the past month a link was established between the Electrical Engineering Group at the University of Belfast - extending from Belfast to Daresbury Laboratory, and thence into the trunk network at Manchester. The Belfast Groups calculations are being performed on the DEC 10 computer at Edinburgh and the resulting silicon-chip geometry is output at Chilton via the FR 80 microfilm recorder.
Up to now, nine Special Interest Groups (SIGs) have been formed among university and polytechnic research workers to advise on the specification of application packages in the following areas:
A considerable programming effort is needed, since it is planned to invest about £100,000 in the software recommended by each SIG. Some of this work will be done at Chilton and some at university and polytechnic departments under contract.
There are many other areas at present not covered by a SIG - eg, crystal structure calculation and problems of chemical technology. In the next few months there will also appear more users concerned with the geometrical design of microcircuit masks for use with the Electron Beam Lithographic Facility, whose hardware is presently being set up in building R1.
The presence of the ICF has already brought together several workers in a variety of disciplines who had previously been working independently. It is hoped that this possibility to pool resources and ideas will be an added stimulus to creative research. It is hoped that some of these projects, particularly in engineering will be to the eventual benefit of British Industry.
(We plan to bring you details of the User applications of the ICF in a future issue of the Bulletin. We thank Peter Davey, Head of the ICF, for the information contained in this report).
