Presented to RAL Division Heads Committee
The organisational structure for the Division is given as an Appendix. The major projects in progress at the moment are:-
Several smaller projects are worked on by the Division and these will be mentioned later.
The Division has two SPSOs apart from the Division Head. P G Davey is responsible for ICF while J E Hailstone is responsible for DCS, Grants and Awards and Database activities. These projects are smaller than ICF so J E Hailstone also contributes to ICF by monitoring performance and user satisfaction as an additional function.
The ASR Board STARLINK project is in an early stage of development. Most of the work so far has been done by C J Pavelin.
The Interactive Computing Facility is now an essential part of the university engineering environment. The size and scope of its hardware and software facilities are such that it is providing a significant contribution to engineering research. A full review of the progress made so far and a revised FYFL was presented to the Engineering Board (EB78-62 dated 14 December 1978) and Council, by P G Davey. The revised FYFL recommended that the hardware purchase of multi-user mini systems should be accelerated and completed by 1980/81. This was accepted and has meant a much higher installation programme in the current year.
The machines associated with ICF fall into three classes:
The complete set of hardware associated with ICF is:
| Site | Computer | Potential User Population |
SRC Share (%) |
|---|---|---|---|
| Edinburgh | DEC 10 KI | 250 | 100 |
| UMIST | DEC 10 KI | 200 | 100 |
| Upgraded Mini Systems | |||
| Sheffield | INTERDATA 8/32 | 100 | 45 |
| Swansea | PDP11/40 | 50 | 75 |
| Oxford | PDP11/45 | 50 | 50 |
| Southampton | PDP11/45 | 60 | 67 |
| Leeds | PDP11/70 | 80 | 60 |
| Manchester Graphics Unit | PDP11/45 | 30 | 10 |
| Nottingham | PRIME 400 | 60 | 80 |
| Multi User Minis | |||
| Chilton | PRIME 400 | 100 | 100 |
| Chilton | PRIME 400 | 100 | 100 |
| Chilton | GEC 4070 | 60 | 100 |
| Bristol | GEC 4070 | 60 | 80 |
| Cambridge | GEC 4070 | 60 | 80 |
| Glasgow | GEC 4070 | 60 | 80 |
| Newcastle | GEC 4070 | 60 | 80 |
| Cranfield | GEC 4070 | 60 | 80 |
| East Anglia | PRIME 400 | 80 | 80 |
| To Be Installed | |||
| City | PRIME 550 | 100 | 80 |
| Sussex | PRIME 550 | 100 | 80 |
| Bradford | GEC 4085 | 60 | 80 |
| Cardiff | GEC 4085 | 60 | 80 |
| Birmingham | GEC 4085 | 60 | 80 |
| Possible Future Installations | |||
| Surrey | PRIME 550 | 100 | 80 |
The two DEC10 KI systems are fully loaded and, in the case of Edinburgh, overloaded. This is particularly worrying as most of the SRC AI community use this machine and are finding it difficult to get adequate turnround. ICF is currently evaluating what should be done in the future. A survey of user requirements is currently in progress and the ICFC intend to decide on the future plans for the two sites in the autumn.
The upgraded mini systems have been variable in terms of their contribution to SRC interactive facilities. The Sheffield Interdata 8/32 has probably the least relevance to ICF's future programme and it is planned to reduce involvement here to nil over a period of a few years. Leeds and Southampton have both successfully provided a service for a large user community.
The new MUM programme started with the purchase of a PRIME 400 and GEC 4070 at Chilton for evaluation purposes. As a result of an assessment involving a collaboration with a number of user groups in universities, it was found that both systems could provide a satisfactory interactive service in a relatively undemanding environment, but in a demanding environment, the PRIME 400 with its extra processing power was to be preferred.
The installation of MUMs at university sites started in the autumn of last year and five 4070 systems have been installed between then and now. This has been a demanding period with site preparation, installation and user training being supervised by P E Bryant. The GEC 4070 at Chilton has been primarily used for system development, although it does have a significant user population as well.
The PRIME 400 at Chilton started giving a user service about six months after delivery and has been overloaded for some time. In a joint arrangement with EBL, a second processor has been installed and the dual system will eventually provide a user service for two hundred people.
Software developments have concentrated on the PRIME 400 and GEC 4070s but with applications software development also being done on the DEC10s and the 360/195. Both the PRIME 400 and GEC 4070 had many important items of systems software and utilities either not provided or provided inadequately when the Chilton systems were delivered. Much of the software effort initially was to remedy this situation. On both systems, the manufacturers' operating systems were also under heavy development which made the job more difficult.
Progress on software development has been slower than we would have liked. For long periods we have had no more than one systems programmer per machine and never have we had more than three. Most of the systems staff has worked long and hard over the last two years.
The broad policy of ICF is determined by ICFC, the Interactive Computing Facility Committee. This body directly approves capital items up to £50K and recommends items requiring Board or Council approval. G Manning, F R A Hopgood and P G Davey attend as members, while C J Pavelin is the only secretary.
The internal management is through regular meetings of PMICF (Progress Meeting for ICF) chaired by P G Davey and held approximately every three weeks.
The management of the Edinburgh and UMIST facilities is carried out by staff of the respective institutions under a contract from Rutherford and joint management meetings (JEUM - Joint Edinburgh/UMIST Management), chaired by P G Davey, are held at three monthly intervals.
To assist in formulating a programme of work and in defining the application software required in particular application areas, a number of SIGs (Special Interest Groups) have been set up by ICFC. These consist mainly of leading research workers who report their requirements back to ICFC, which accepts or modifies their findings; Where programmes of work are defined, either the SIG or a smaller Steering Group continues to vet progress. SIGs exist for the following areas:
The Division provides a secretary for the various SIGs and aids them in formulating their plans.
To keep the user population informed of significant events, a Newsletter called Rapid Response is issued at regular intervals. Nearly 2000 copies of each issue are distributed.
User Committees have been set up for the Edinburgh, UMIST and PRIME services.
Management of Edinburgh, UMIST and the MUM sites (five at present) on a daily basis is delegated to the universities concerned. This is done by means of a three-year rolling contract, under which each site agrees to provide a service in return for quarterly payments according to a budget negotiated each year. Management meetings with the MUM sites are held several times a year, including one to agree the next year's budget and another to reconcile actual expenditure over the previous year.
The ICF is run as a fully allocated facility. The consequence is that considerable effort (3 my at present and increasing) is devoted to authorisation of access and resource control. The total number of users presently authorised to use the system is about 1100, expected to rise eventually to 1700. Detailed comments are made on some 300 grant applications per annum, often leading to technical discussions with applicants.
ICF support a large proportion of the users accessing facilities provided by the Computing Divisions. There are currently about 1000 ELECTRIC users compared with 1100 ICF users. The ELECTRIC user gets about 2 hours connect time per week compared with 5 hours for the ICF user.
The ICF group is divided into four sections.
The structure is:
The PRIME systems group has an ongoing need for about one man year of effort in systems maintenance (installing new systems, minor changes, finding bugs, etc). P J Newton, who is responsible for this, has also provided tools for monitoring systems performance and is attempting to pinpoint bottlenecks. L O Ford who is in charge of the systems section has been dealing with the problems of connecting the two PRIMEs, of networking PRIMEs in general and connecting them to a range of other machines. The group is preparing for the installation and remote maintenance of further ICF PRIMEs. In addition to a range of minor enhancements, an incremental dumper is being implemented with the aid of two students.
L O Ford is taking a year's leave of absence in the summer, to visit a University Group in the USA. His place is being taken by G W Robinson, who has been responsible for the DECNET network between the two DEC10s and the local DN82, for HASP on the PRIME and File Transfer Protocol on the GEC.
M P Keane handles user support single handedly, although with a user population of well over 100, the support is limited and the manual produced by the Division for PRIME Users is now well out of date. It is recognised that the support levels are inadequate to provide a first class service.
Rapid Response, the newsletter of ICF, is edited by P Smith; it has a circulation now of about 2000 and there are four issues a year. P F Smith also runs the Special Interest Group in Computer Aided Architectural Design.
With a user population of over 100, the PRIIME support staff are very active. The Division has produced a User Manual to supplement those provided by the manufacturer. A User Group is in the process of being formed.
The function of the Finite Element (FE) group is to implement the recommendations of the two FE Special Interest Groups. (The work is now co-ordinated by a smaller FE Steering Group.) Following a recently completed evaluation exercise, two FE packages (ASAS and BERSAFE) are being supported. A two level FE Library, consisting of building blocks needed by FE code developers and examples of how to use the building blocks, is currently under development and a first release should be available by September, when an FE User Meeting is to be held. In addition, work is proceeding on the provision of interactive graphics pre- and post-processors to the FE software.
The structure is:
The major work over the last few months has been the installation of GEC 4070s at remote sites. So far, five machines have been installed and unfortunately deliveries have bunched. This has stretched the resources to breaking point and has involved very long hours by some people and it has only been with considerable goodwill that installations have succeeded. Nonetheless, the exercise has been exceedingly successful. All the computers have been installed in three or at most four working days and thereafter have provided full 24-hour service, 7 days a week. In general, it takes six months to prepare a room and this is done with local effort. Co-operation between the sites and SRC has been very good.
The Hardware and Software has been very reliable, although the work load on the machines is not yet heavy. In fact the build up of work has been steady but slower than expected. The Benson Plotters attached to each machine have proved a good buy and have so far been faultless in operation.
The sites are part of the SRC network. Several machines have been unable to participate at an early date due to lack of lines. This has been unfortunate but not a disaster. It has made support difficult since this tends to be via the network. All sites now have lines although the full networking system is only just reaching perfection when its full potential will be felt. Its full potential provides interactive use of any GEC 4070 from any other, as well as access to/from and via the 360/195. Files may be transferred between the GEC 4070, and jobs may be sent to the 360/195 or Daresbury site.
The machines are supported from Rutherford Laboratory by five people, who are: Paul Bryant (managing the project), D C Toll (a systems programmer), J R Gallop and J J Hutchinson in a support role and part of an operator - R Brandwood. In addition, each site has one man year of effort on operations and support. Thus, currently about ten people support the six machines and will eventually support nine. This level is too small, considering they also undertake installation. In addition, the project is vulnerable to resignations, since there is no effort available for back-up. Nonetheless, the philosophy of running the machines in this way is now confirmed as very economic.
Support tends to be via the network. Users complain by typing their complaints and these are looked at by RL staff at least once a day and replied to. This is proving satisfactory and quite popular. It is fostering a community spirit amongst the machines which is very valuable but unquantifiable. Basic Software, like operating systems and compilers, are always provided by RL. The aim is to supply these via the network. So far most software has been taken to sites by hand, but experiments at Bristol have proved the feasibility of delivering by the network, and this will be done in future. In fact it will be deemed a failure if anyone has to visit a site other than for political reasons. This will save vast amounts of travel time as well as petrol, which is a point worth making with the increased problem of travel.
Great changes have been made to GEC software with the aim of making the machines as simple as possible to operate while maintaining good control of them. For example, on an IPL the operator is forced to give a reason for the IPL and the date and time, everything else is automatic. New SRC users can only be entered by RL staff but new university users can be put in by university staff. This is convenient, since the allocation of SRC job numbers is done at RL. Another example of a development has been the automatic speed selection on terminal ports. This has allowed the site to connect any terminal to any port without reconfiguring software or any other operations.
In general site are very happy with the GEC machines. Contrary to expectation, they have so far not been troubled by the speed of the machine or the 16K problem.
Despite the problem with lack of staff, uncertain and late deliveries of GEC machines and the problems of SRC committees, the project has the sweet smell of success.
The structure is:
The graphics group is divided into two sections. The first, under D C Sutcliffe, is responsible for providing graphics packages on the PRIME and GEC computers. This involves maintaining the two major packages, GINO and FINGS, providing device drivers for the many different devices available on these machines. In particular, special attention has been given to black and white and colour displays marketed by SIGMA Ltd and close liaison with the company's developments is maintained. In addition, higher level packages dealing with contouring, 3-D histograms etc. are being mounted. This work is being done by M R Geary, who joined the project in January of this year. The section is also involved in the NAG graphics library project, and the advances in raster graphics terminals and techniques.
The second section is involved in writing basic software for the FR80 microfilm recorder. H K F Yeung has been working on DRIVER for some eighteen months. The project was started nearly three years ago. Delays in completion have been partly due to changes in personnel involved and partly due to a long enforced absence due to illness. Now, however, the first release has been introduced (June) and the end of the first phase of the project is in sight.
The DACD group, led by P A Dewar, is in the process of implementing the findings of two special interest groups (analogue circuit design and digital circuit design) which presented their reports to ICFC at the end of last year. ICFC approved a staff of six initially, but it is proving very difficult indeed to fill the remaining two posts. In some way, circuit design is still in its computing infancy, as far as academics are concerned and the packages that are available tend to be large, disjoint and each serves one small area of the whole subject. Most of the important software is currently owned by industry, who are very jealous of its confidentiality. Nevertheless, some headway is being made.
I Benest joined the project last August. He is involved with analogue circuit design; in particular, the graphical representation of circuits and the user interface. He is currently producing a demonstration system on the UMIST DEC10, as a basis for ideas towards a fully functional design system. Other analogue analysis tools mounted include ASTAP, SPICE and NAP, all on the 360/195. Dr R Gay, who joined in April, is working on the comparison between these packages.
J McLean is concerned with digital circuit design. He joined the team in January and has been looking at the Digital Design Language system and ISPS, both mounted on the UMIST DEC10. It is hoped, in the long term, to base the digital design system around a chip library system that formed the basis of J McLean's PhD. This is being hampered currently by the lack of suitable system software to complete the job.
Further work in this area is being done by the use of Extra Mural Development Contracts with universities, notably Essex and Reading. The group is also closely involved with Instrumentation Division (wire-wrap program) and Technology Division (GAELIC), since ICF users require access to these facilities.
As well as his involvement in all the above areas, P A Dewar is concerned with microprocessor software, with the provision of cross-assemblers and simulators. Work is just beginning in this area.
A steering committee has recently been formed to monitor the progress on behalf of ICF.
A major problem here and in other applications areas is the need for specialised systems software to provide the basis for the application system. Currently, ICF has insufficient systems people to provide what is required and consequently the applications work proceeds at a slower pace than would be desirable.
The AP120B is a peripheral, currently attached to the PRIME. It provides fast vector arithmetic facilities to suitable programs, giving speed increases of up to 100 times that obtained on the PRIME alone. Dr Bryden is half way through a year's project involved in evaluating the usefulness of this device in the various disciplines covered by the ICF. This involves liaising with potential users, learning how to make the best use of the device and implementing the manufacturer's software. Early results indicate that users will need to put in a lot of work before being able to get the full benefit from such a device. Once they do, however, it becomes a very cost-effective method of providing computing power.
Dr Jezek is here for a year from Czechoslovakia under an exchange scheme. His interests are in numerical and compiler techniques and he is looking into ways in which the AP120B might be used to help solve equations met in Finite Element work. In addition, he is investigating ways in which the special FORTRAN compiler might be improved.
The people involved are:
The ICF Management Section has three major functions:
The process of applications assessment begins with the initial enquiry by the potential user with a project which may attract SRC support in the form of a Research Grant. If the user has no experience of interactive computing or requires a short term feasibility study, a period of pump priming may be arranged. Enquiries are made direct to RL or via the external host managements. Potential paying customers are also dealt with by the Section.
When applications are being prepared for submission to SRC, advice is given on resources required, facilities that can be provided, terminals available for loan, software supported by ICF. This activity is carried out by the Section with technical support from the other sections of ICF.
After submission of the application to SRC Central Office, a request for comments is received and these are returned to the appropriate committee for inclusion in the approval mechanism. Other applications with computing content, but not directly requesting use of ICF are assessed for relevance to ICF and post-approval discussions with the applicant may arise before an announcement. The most recent round of grant applications produced more than 90 proposals in total. The comments on these were written by J Brown and B G Loach in the main.
After approval and announcement of grants, the authorisation of ICF facilities is handled entirely by the section which prepares and distributes the ICF authorisation sheet to the grant holder, host computer management and communication group, where additional hardware facilities are required. Loan of terminal facilities is handled by the Section.
External ICF Finance
Each ICF external service is financed through a recurrent support agreement with RL. The basic contract consists of a rolling three or four year agreement with an annually agreed operating budget.
Payments are generally made three months in advance and cover staff, equipment, maintenance, consumables and overhead services. All agreements are subject to detailed annual accounting and payments are adjusted in retrospect.
Minor capital provision for external ICF services is handled by the Section and includes provision of enhancements to communications equipment replacement and enhancements of peripherals and purchase of software.
All usage accounting is collected by the Section in order to provide regular ICF wide statistical information. Common accounting systems for all types of computer in the ICF are under development.
ICF Terminal Pool
The ICF terminal pool consists of more than three hundred devices covering a wide range of types. Of these, two hundred are on a single maintenance contract with a nationwide firm, Cable and Wireless. When loans are authorised, the terminals are checked, packaged and dispatched to users who are informed of maintenance arrangements. Comprehensive information on the state of the pool is maintained at all times. Regular meetings are held with Cable and Wireless to ensure satisfactory maintenance arrangements for all types of terminal.
New terminals are purchased for the pool by the Section. Statistical information on the use of terminals is collected.
Special needs are catered for by one-off purchases and new types are examined regularly.
All aspects of the work of the Section would be considerably assisted by the availability of a computerised management information system across the network. No progress has been made with the aspect due to lack of staff effort.
The current Forward Look (1978 prices) shows capital expenditure as follows:
| 1980/81 £K |
1981/82 £K |
1982/83 £K |
1983/84 £K |
1984/85 £K |
|
|---|---|---|---|---|---|
| Multi User Mini | 510 | 170 | 0 | 0 | 0 |
| Other | 290 | 327 | 358 | 336 | 290 |
| TOTAL | 800 | 497 | 358 | 336 | 290 |
Roughly half of the second line provides for a systematic programme of upgrading the MUMs installed between 1978 and 1981, adding more memory and increasing processor power at individual sites where the user pressure is greatest. The remaining half is mainly required for array processors and single-user machines both for assessment and for providing user services.
Recurrent costs decrease slightly from £1653K in 1980/81 to £1416K in 1984/85. The latter figure breaks down as follows:
| £K | |
|---|---|
| Bare maintenance of MUMs and terminals at university sites | 375 |
| Support of larger installations at Edinburgh and UMIST (may be reduced) | 360 |
| Chilton service, managing and operating the network, central systems/operations staff (16 direct my) | 379 |
| Centrally supported applications software purchases, and applications development staff (18 direct my) | 302 |
| TOTAL | 1416 |
SRC Central Office would prefer the first two items (at least) to be taken over by Universities' own funds. Discussions are in progress but the ICF view is that this solution is unlikely to be viable.
The MUM network achieved by 1981 will be a facility having enormous potential. To realise this full potential for the benefit of SRC grant holders will be an exciting and demanding task requiring the best available programming staff to acquire, document, and where necessary create applications software. A considerable part of this work can be done under contracts with University and Polytechnic departments but experience has shown that careful and determined management is required if this is to be successful. Applications software will mainly take the form of a few "black box" packages together with many easily-interfaced routines available in a library. Documentation and support of packages and libraries must be of a high standard since research workers cannot be forced, only tempted, to use them.
Provision of good interactive computing beyond 1985 demands advanced work on new hardware facilities. These may well be based on very powerful, cheap single-user processors with fast links to local centres having specialised equipment for file storage and exotic peripherals. The work underway in the Division in support of the Distributed Computing Systems programme of CSC will be invaluable here. On the basis of development work at Chilton and by DCS sites in the Universities, the ICF should decide by 1981 on a preferred technology, then concentrate on extracting this from the R and D field, conduct an assessment involving real users, and finally make costed proposals for the implementation of a new facility which could eventually replace the MUM network. This should be thought of as a means of providing an interactive interface available to the majority of users of all SRC's central computing facilities within the next ten years.
The structure is/was:
This project was set up to implement replacements for the SRC Grants and Awards administrative systems on the ICL 1906A when that computer was shut down. The 1906A systems were batch oriented; were designed, implemented and maintained at any one time by up to ten programmers from three different software houses. The grants system was started in around 1972 and evolved continuously until mid-1978. It involved about one hundred and fifty separate computer programs. The Awards System now in full production was begun in 1975.
The 2904 computer is much smaller than the 1906A, with far fewer program development facilities, but it is dedicated to this application. The major requirement of the new system(s) was that update and enquiry of the database should be interactive, via VDUs distributed around SRC Central Office at Swindon.
All the work done on the new system has been carried out in close collaboration with Swindon Office staff.
This has been completely re-written, but is able if required, to preserve the same interface to the users as offered by the 1906A system. Fully interactive update and enquiry are available. Two commercial software products have been employed in the programming: TPS teleprocessing software and the TOTAL Database Management System. The latter should facilitate maintenance and future developments of the system. Three and a half RL staff and one from Swindon Office were involved in the design, programming, system maintenance and day-to-day administration. Some short term assistance with the programming was provided by contract programmers.
The 2904 Grants System went live and replaced the 1906 system on 25 June 1979.
Although preliminary work on the 2904 version has commenced, this is really awaiting a positive decision whether to convert all or part of the 1906A system or to start again from scratch. The favoured solution is a complete re-specification to enable full exploitation of the facilities offered by TOTAL and TPS.
Once the directive is received, the implementation of the Awards System should take about two years. Concurrently, various enhancements to the Grants System are envisaged; subject classification and automatic indexation of staff salaries are two examples already being worked on. Much more use could be made of other RL computing facilities e.g. the FR80 (to produce print masters for the Yellow Book and other publications) and FAMULUS (to produce KWIC indexes of Grant Titles). These two exercises would be particularly eye-catching and may earn some recognition for the efforts of the staff working on the 2904.
The staff working on this project are subjected to particularly strong pressures in the following areas:
The majority of the users of the system are not computer people. Many are not familiar even with SRC office procedure. Their reaction tends to be vociferous if things go wrong. They also have difficulty in understanding why apparently trivial requirements need very explicit specifications and may take the programming staff much effort to implement. Consequently, the deadlines set are not always realistic.
There are groups within Swindon Office who have different views on office procedures, how the 2904 computer should be used, administered and located, whether it is even needed.
Despite this, the staff involved have between them produced some 12,000 lines of COBOL code driving both TPS and TOTAL. The system has been proved by parallel runs on test data and copies of the real database. The critical exercise of implementing a program to convert the 1906 serial files to an error-free network data base (and all that this implies) has also been carried out successfully.
The SRC's Specially Promoted Programme of research into Distributed Computing Systems is run by a Panel of the Computing Science Committee of the Engineering Board and coordinated by an SRC Coordinator and an Industrial Coordinator. The SRC Coordinator is responsible for ensuring co-ordination between the groups while the Industrial Coordinator's main responsibility is to ensure that industry is aware of the research in universities and makes contact with them where appropriate.
The initial Coordinators were myself and Gill Ringland of Cap with R W Witty acting as Technical Secretary and Swindon Office providing the Panel Secretary.
Since April 1979, R W Witty has taken over as SRC Coordinator; J Tucker of Logica Ltd has become Industrial Secretary. Swindon Office provide minimal support for the Panel and most of these duties are being forced on to the laboratory.
The Coordinator is responsible for talking to potential grant applicants, modifying grant proposals where possible to plug holes in the programme and present grant applications at Panel and Committee meetings.
The Coordinator visits the groups involved in the programme twice a year and organises meetings for groups with similar interests and organises workshops and conferences.
The number of groups involved in the DCS Programme is 32 and the total value of the grants is £2.3 million.
Major meetings and workshops organised in the last year were:
| May 1978 | Real Time | York | SIG |
| May 1978 | Ironman | York | Workshop |
| June 1978 | Dataflow | Newcastle | Workshop |
| Aug 1978 | Dataflow | RL | SIG |
| Sept 1978 | Languages | Warwick | Workshop |
| Jan 1979 | Dijkstra | Oxford | Workshop |
| Jan 1979 | Academic DCS | IEE | Colloquium |
| Apr 1979 | Networks | UCL | SIG |
| May 1979 | Industrial DCS | IEE | Colloquium |
Many of the Grant Proposals are concerned with developing systems involving a number of processors cooperating to perform a specific task. Some research programs would like to use ten or more processors for certain periods to do full tests of particular systems. However, they only require a few processors during program development.
The Panel were awarded funds in 1978/1979 to provide a pool of processors and to provide equipment to aid communication between the research groups.
The Pool currently contains:
10 Newbury VDUs 6 EMI Magnetic Tape Decks 7 Diablo printers 10 LSI-11 Computers with dual floppy disks
To aid communication between the groups, an EMR contract is in progress with Belfast University to provide a portable PASCAL PLUS compiler which could be used by many of the groups as the standard concurrent programming language.
Another EMR agreement is to be placed with the University of York to provide X25 communication software between PDP11 systems using the UNIX System.
To provide a central database of information related to DCS grants and to provide a facility for research groups without PDP11 UNIX systems, a small PDP11/34 system running UNIX has been installed at the Laboratory and two complement places exist for programmers associated with the system. Jobs have been offered to suitable candidates interviewed recently and one has accepted so far.
A further capital provision of £163K has been made for the current year and the pool will eventually contain:
10 VDUs 11 Diablo Printers 10 Magnetic tape decks 15 LSI-11 computers 7 Personal Computer Systems - either TERAK or Three Rivers PERQ.
Administration of the pool is combined with the ICF terminal pool. A File Transfer Protocol is being implemented at Chilton to allow files to be transferred between LSI-11s and other computers.
The SRC Coordinator is responsible for the purchasing, distribution and maintenance of this pool of equipment which will be worth £340K by the end of this year. The SRC Coordinator is also responsible for the placing and supervision of EMR contracts and the running of UNIX system and staff.
A panel of the Astronomy II Committee of the ASR Board has recommended the purchase of six multi-user mini computer systems for interactive image processing for astronomers. The computers will have colour raster displays attached, and will be linked in a star network, with the centre at Rutherford Laboratory. The other nodes are to be at RGO, ROE, UCL, Manchester and Cambridge. The project, known as STARLINK, is to be managed by Atlas Division. The Division (C J Pavelin) was responsible for the computer tender, RGO (K Hartley) for the Image Display Tender. The purchase of six computers with a total power exceeding a 360/195, 5000 Mbytes of disc space and 12 Mbytes memory, is being recommended. The total value, including image displays, is £1.7M.
It is expected that the project will run as an adjunct to the ICF, but will have its own head, probably appointed from the astronomical community.
Some small projects exist in the Division and some effort is put into areas that are seen to be growth areas in the future. These are:
Work continues on the production of volumes in the Molecular Structures and Dimension Bibliography - Organic and Organometallic Crystal Structures Series. Mrs K Crennell has worked for some years in close collaboration with the Crystallographic Data Centre, Cambridge and this work continues and is highly regarded.
Other work in text processing includes the preparation of many different fonts and the mixing of scientific texts and diagrams. Advice to the Laboratory includes index pre-direction, catalogues, etc.
Because of the importance of documentation for computer users, word processing systems are being investigated with a view to integration with main computer systems.
University and RL workers are increasingly aware of the need to use computer based information handling systems. An increased effort is being devoted to the task of making such systems available and in advising and instructing users of the appropriate systems for their work and in the techniques required to implement their data projects.
Mrs A M Walter and Mrs J 0 Lay are working in this area under the general supervision of M F Chiu.
Demand for this work arises principally through the ICF but other RL projects have expressed interest and are being helped to penetrate this new technology. Two major systems have been mounted on the ICF PRIME 400:
A number of the staff contribute to either national or international activities in the computing area.
K Robinson and D C Sutcliffe have been involved in the development of the NAG Subroutine library. The Division is partly responsible for the production and verification of two chapters in the NAG Library - Summation of Series/Fourier Transforms and Graphics.
D A Duce has been a member of the BSI Graphics Group that has been active in putting the UK view forward with respect to graphics standard proposals.
R E Thomas is a member of the GINO-F Technical Committee which is responsible for vetting proposed changes to GINO-F and advising the CAD Centre of worthwhile changes. He is also a member of the Group involved in defining the DAP FORTRAN language.
R W Witty is the UK member of the Organising Committee for the First International Conference on Distributed Computing to be held in Hunstville, Alabama and is responsible for the UK refereeing for the Conference.
P E Bryant has been active for a number of years as Technical Officer for ACTP on several projects including ICL CAFS, ICL DAP etc.
D C Toll is currently Technical Officer for the DEMOS system being developed at NPL.
F R A Hopgood is a UK representative to the ISO Working Group associated with Graphics Standards and is Co-Chairman of the Editorial Board which is currently assessing the three Standard Proposals from Germany, Norway and USA. He is also a member of IFIP WG5.2 - Graphics Committee and has been on the Organising Committee for the two major graphic workshops in the last few years, Seillac I (1976) and Seillac II (1979). He also has a full Professorial post at Brunel University in the Department of Computer Science (10% of his time).
C J Pavelin is Secretary of the PRIME European User Group.
J O Lay is a member of the BCS Study Group in Computers in Surveys and # maintains contact with developments in statistical tabulation software.
P G Davey is a committee member of the British Pattern Recognition Association affiliated to the International Joint Conference on Pattern Recognition.
L 0 Ford is a member of the Management Committee for the Manchester University Graphics Unit.
P Kent continues to participate in International Computer Chess discussions and expects to contribute to the European tournaments and IFIP Conference this year.
The major projects of the Division will continue for the next two. years at least. The emphasis within ICF is likely to. change from systems development to management of the facility, application program development and support.
Both the DCS programme and the ICF see an increasing emphasis on powerful personal computer systems being the main user interface computer by the end of the decade. Such systems, which are now appearing on the market, will have high quality raster graphics as the main output device and both operating system and compilers implemented in firmware. Resources, such as file storage, powerful processing facilities, databases etc. will be provided by local area networks with servers providing resources. It is important, therefore, that the Computing Division is actively involved in research in these areas. The personal computer side will arise naturally from the ICF applications side and the DCS programme. The local network research is mainly being done by C & A Division at the moment and will be of major importance to. site computing development.
The STARLINK facility will provide the astronomers with image processing facilities at least comparable with their overseas counterparts. Parallel processing is particularly applicable in this area and development will continue on both the AP-120B and DAP. It is important to realise that the volumes of data required make it impractical to transmit over communications networks either now or in the foreseeable future. The CRAY-l at Daresbury is unlikely to be of any use in this area, therefore.
The interactive applications areas continue to require large batch facilities with fast turn round. The ICFC is anxious for the CRAY-l to be available to ICF users.
The inability to recruit staff means that everything needs to be done to improve the productivity of those we have. Work will continue on the development of software tools for program design, writing and testing.
There are two major problems in the Division at the moment.
Four people were either nominated or boarded for promotion this year and yet none were successful. Their view is that insufficient weight has been given to either the importance of the project, the market forces associated with their expertise, or the magnitude of the data processing task.
