1. The Director Laboratories, Dr A E Hughes, was asked by Council to produce a plan for its scientific establishments into the 1990s, set against the general policies expressed in Council's second Corporate Plan (1989). The Corporate Plan guideline for the work at the laboratories is as follows.
The main purpose of Council's establishments is to provide and develop advanced research facilities and services for academic research which cannot be operated on the university site because of their size or complexity.
2. The terms of reference given to Dr Hughes were:
3. Dr Hughes presented his interim report to Council in October 1989, there was a wide-ranging discussion and the points made will be taken into account into producing a definitive report to Council in the Spring or early Summer of 1990.
4. In introducing his report Dr Hughes made a number of points including:
5. There was little coherent thread running through the discussion. A number of Council Members, particularly those who are connected with programmes that depend very much on the laboratories were highly supportive. Some Council Members were critical and would wish to see more activity in HEIs at the expense of the labs. The message that came through was that partnership is the crucial concept. In all our dealings with HEI scientists and engineers we need to emphasise our willingness for work to be carried out in the most appropriate place. Sometimes this will be in an HEI sometimes it will be in the Lab. The decision on the balance between these two complimentary routes must be arrived at in partnership with the HEI community. SERC's laboratories are at their most vulnerable when they are thought to be taking programmes out of the HEI sector for no good reason. The laboratories are strongest when they are seen as partners and collaborators whose principal aim is to make HEI research easier to do and more effective.
6. The annex contains the contents page and those sections of the report, chapters three and six, that relate most directly to RAL and paragraph 9.11. Dr Hughes concluding remarks. Associate Directors have copies of the full report.
RAL has the most diverse programme of all the laboratories, being very roughly equally committed to each of the four SERC Boards and to centrally-funded activities. Each of the main Board-funded programmes and infrastructural aspects of the Laboratory are covered in the following sections. The main centrally-funded activity is the provision of mainframe and supercomputing facilities and services, including, in the case of the. Cray XMP-48, the provision of a facility for all the Research Councils, coordinated with the two major university supercomputer centres at Manchester and London supported by the Computer Board. An important part of these functions covers networking and the responsibility for JANET, which is part-funded by the Computer Board. Since central computing is a federal activity, and includes some work carried out at Daresbury Laboratory, it is discussed separately in Section 6.
RAL also provides the Council Works Unit and some administrative support to Swindon Office. These are not explicitly addressed in this Report.
The primary role of RAL in particle physics will continue to be the support of the UK particle physics community by the provision of large scale facilities and services that cannot be operated on a University site or which are more appropriately sited centrally .
The form this support will take will evolve as demand dictates - RAL has dramatically changed its role in the past to accommodate the needs of the community: from an accelerator laboratory, to a laboratory supporting a large number of relatively small experiments, to a laboratory pivotal in the provision and support of a relatively small number of very large experiments at overseas laboratories. The 1989 Corporate Plan proposes that funding of Nuclear Physics should stabilise following the planned reductions of recent years.
Although the world particle physics scene in the mid/late 1990's is unclear certain features are already known. Large and long term experiments at overseas laboratories will still dominate the field; however these will very likely be mixed with smaller, but not small, experiments in underground locations. Almost without exception these experiments will involve international collaborations. In the UK, the resources available to the particle physics community are under great strain and will probably not greatly change for the better. This makes it imperative that the UK effort should be as well coordinated and as efficiently run as possible. RAL's role in this will be crucial. The long term support of the programme by RAL groups of physicists, engineers and technicians will ensure that UK responsibilities are honoured, that duplication of effort is avoided and that UK academics can participate fully, even if, as often happens, they cannot spend long periods away from their own institution. The coherence this gives to the UK effort in particle physics has been commented on favourably by scientists in countries where this focus is lacking.
RAL will act as a centre for the UK community bringing together diverse technologies, disciplines and groups from all over the UK. These are likely to be in both traditional areas of engineering and accelerator research and in newer areas of microelectronics, specialised detectors, parallel computing, data acquisition systems and software engineering. The laboratory environment, with its differing scientific and technical objectives, provides the synergy needed to tackle new challenges.
Computing and communications will play an ever increasing role in the future. RAL will support both through its computer and particle physics departments. Close support in specialist areas such as networking, data bases, graphics, on-line systems as well as applications programming will be increasingly important and RAL will supply this expertise to the community both directly through collaboration and through the central computing department.
RAL will seek to maintain its technical and scientific expertise in detector development and accelerator technology by tendering for appropriate R&D_ contracts for new accelerators or other high technology equipment from overseas laboratories.
New science will bring new opportunities and RAL will be in a position to exploit these. It will use its expertise and large central facilities to coordinate and, where needed, initiate UK programmes at accelerators and other sites around the world.
There is likely to be a growth of interest in topics that do not involve accelerator-based experiments and this could lead to the setting up of new interdisciplinary institutes whose remits are somewhat different from IRCs. The US model of the Particle and Astrophysics Institute set up at Berkeley may be an appropriate one. RAL would be an ideal host for such institute with its diverse expertise and its traditional role of support for HEI groups, together with its proximity to a major research activity at Oxford.
Apart from the support role, RAL physicists, computer specialists and engineers will, as in the past, be uniquely qualified to make their own research contributions to the field; this will almost always be in collaboration with HEI colleagues.
A close link between experiment and theory will continue to be essential in making progress in particle physics. The community has reviewed the particle physics theory group at RAL twice in recent years and has concluded that its excellence and expertise in phenomenology are not only much needed and not found adequately among the interests of HEI particle physics theory groups, but also that the role of the group in the wider activity justified its continuation at RAL. In principle it would be possible to provide the UK particle physics community with expertise in phenomenological theory from an HEI group. However it would be counterproductive and dispiriting to repeat these reviews in the immediate future, especially when there is no evidence of an HEI Department that is particularly interested in taking over the responsibility. Some further brief comments are made on this in Section 9.3.
RAL will be as important in the 1990s to the UK particle physics programme as it is now. Its role will change somewhat and the mix of skills needed will also change. However it is uniquely placed to respond to these needs. RAL acts as a focus and a unifying force in the community, making the administration and organisation of UK participation in experiments transparent, responsive and effective within the resources available. Its role will remain important in the 1990s as the scale and sophistication of experiments continues to increase.
RAL provides a central and coordinating role for Council's support of space science and astronomy, particularly for major space astrophysics and earth observation instrumentation. The Laboratory also has responsibilities for parts of the ground-based programme, notably in geophysics, and for Starlink. Many of these activities are substantial in scale and in the number of collaborations and partners whose contributions need managing and coordinating to ensure that the projects lead to a successful conclusion. RAL's ability to provide interdisciplinary teams of scientists and engineers, covering managerial and technical expertise and including some involvement in the research, is a virtue of the Laboratory's presence in this field and can be expected to remain so throughout the 1990s. The skills and reputation of the Laboratory in Earth Observation are recognised as a particular strength of the UK in contributing to the international effort on global environmental research.
By the nature of the long time scales (10 years+) involved in planning space or ground-based projects in astronomy and geophysics, there is already a reasonably clear picture of programmes and possibilities for the APS Board programme in the late 1990's, provided that level funding is maintained as envisaged in the 1989 Corporate Plan. It is assumed that the remit of the Laboratory to support UK academic researchers will remain largely unchanged, and further that the status and role of the British National Space Centre as a co-ordinating body for the national space programme will not significantly change. Compared with the present day three trends can be anticipated:-
Against this background, future trends in the various programme areas can be examined.
The programme base for this planning is that contained in the current SERC Space Plan, as updated by the BNSC Programme Boards recently, and includes such missions as SOHO/CLUSTER, SPECTRUM-X, Lyman, XMM, Cassini, ERS-2 and the Polar Platforms. By the late 1990s the provision of high quality facilities (clean rooms, test facilities etc) and services (management, quality assurance, parts, etc) on a timely basis to HEI-based builders or designers of space hardware will be increasingly important. HEI scientists will be more involved in conceptual design, breadboarding and early development work; flight hardware will increasingly be built only in the few better equipped HEI laboratories, at RAL or in industry. However, easy access will be needed for testing and calibration services by researchers, without the prohibitive costs of using industrial-scale facilities. Facilities at RAL will be concentrated in one large-experimental hall, with support/development laboratories around the periphery of the hall, in close proximity.
The recently published Ground Based Plan highlighted three projects for particular attention: a Very Large Telescope, a Gravitational Wave Observatory, and a Polar Cap Radar. It is likely that by the late 1990s existing radar facilities such as EISCAT in Scandinavia, and the MST Radar in Wales will still be in operation, but will require decreasing RAL support. Attention is very likely to be focussed on the development of a new northern radar in the EISCAT system (the Polar Cap Radar, PCR). The extensive expertise at the Laboratory in advanced radar design and engineering management, demonstrated on many other projects, should be used as part of the UK contribution in the development of the PCR.
It is to be hoped that, by this period, most of the engineering design and development required in support of a gravitational wave observatory installation in Scotland or elsewhere will be complete, and the Laboratory's role will be limited to maintenance support, until and unless an upgrade to the facility might become necessary.
By the late 1990s, however, it is likely that the major development in the ground-based programme will be the next generation of ground-based telescopes, probably a very large (8mm diameter class) optical/IR facility. It is foreseen that the engineering and management skills available at RAL, and tested in previous projects like JCMT, would be an important element of the UK contribution to the project (see also section 5).
The present level of interest in issues of the global environment is likely to stay with us through the decade. The expertise of RAL in the design, development and utilisation of advanced measurement systems for the observation of the Earth's atmosphere and climate is expected to be widely called upon, in space projects through BNSC, ESA and NASA, and other scientific activities. The BNSC has reflected government policy by. declaring Earth Observation (ED) as the priority growth area for the UK space programme, and within the science programme the implications of this are likely to be an increased opportunity for participation in EO missions such as the ESA and NASA Polar Platforms, together with under-pinning scientific R & D in HEIs and research institutes. It should be noted, however, that the resources for making best use of these opportunities have yet to be found.
The need to provide computing, hardware and communications expertise for HEI users, including data links with other organisations and the management of the Starlink facility, is expected to continue through the 1990s. It is likely that the balance will move towards acting as a gateway to national and international data archives; providing centralised support in software, systems skills, and large databases; and serving users with considerable computing power and storage capability at their own sites. Plans are in hand to bring together the various data support activities into one Space Data Facility, where the required support can be more effectively provided.
The data and communication activities of the sort described above are not on a very large scale and the main (and considerable) advantage of concentrating them at RAL is the focus, integrated management, coordination and regular service to the community that this provides. Individually, the activities might in principle be conducted outside the laboratory framework, although it is not clear that their nature is such that they would then be given the dedicated attention that they need. They do not, for instance, give the local operators any particular research or training advantage, such as accrues from having immediate access to an advanced scientific instrument. Starlink itself was subject to an extensive debate in 1986 and a firm decision made to continue its successful operation by RAL. It is doubtful that the totality of these activities could be handled as effectively by an HEI since a substantial skilled resource is required (ea 20 people). Individual items could be considered when they come up for natural review (eg the next review of Starlink), on the basis of seeking costed expressions of interest that would be expertly reviewed. RAL would undoubtedly continue to be a strong contender, whose track record is not in dispute.
In attempting this projection, it has been assumed that SERC resources will remain roughly constant in real terms over the next decade, and that, as stated in the SERC Corporate Plan, the APS Board share of those resources will not fall any further. Within this assumption, a number of projects in both astrophysics and geophysics, in space and on the ground for the late 1990s, are reasonably well identified already. These projects will require support by the skills, expertise and facilities which the Laboratory is likely to maintain through the next decade.
Neutron Facilities are currently subject to review by a panel set up by the Science Board. The plan presented here is therefore subject to the outcome of that review. The plan is to provide for the development of ISIS so that it retains its currently pre-eminent position as the world's most powerful pulsed neutron and muon source by a factor of about five.
ISIS is a young facility which has a scientific programme only three years old. Already it has demonstrated a rapid development in the quality and range of science which can be studied using the ISIS neutrons and muons. Experience has shown that these developments will continue, provided that the source and its instrumentation can be kept at the forefront of the technologically possible. Within the next few years there are improvements which could be made to the facility to increase the overall output of science in a cost-effective manner. The options are to increase the number of neutron detectors on the existing instruments, to provide new spectrometers to make full use of available neutron channels and to develop the accelerator and target station from the present 100 microamp level to the design of 200 microamp. These and their costs have been considered in detail in a recent paper to Science Board and are under discussion to set the immediate courses of action.
During the early years of the 1990s the plan for ISIS will therefore be to carry out such improvements, alongside the exploitation of the facility through operation for users, balancing operation and development within the resources that can be secured from Science Board and overseas collaborators. The level of resources likely to be available will not allow major upgrades, but these must be considered for the second half of the decade if ISIS is to be capable of development into a next-generation world-class facility for neutron beam and other research.
By the late 1990s it would be technically feasible to upgrade the accelerator system by the use of an 800 MeV linac. This would feed the present synchrotron ring modified·to be a storage ring producing the required proton pulse length. 1.6 mA, 8 times the current design performance. The accelerator system would be designed to feed two target stations in a multiplexed fashion. Each target station would be optimised to match a range of neutron scattering spectrometers designed to meet the developing scientific requirements. This would approximately double the number of spectrometers possible with the present target station.
There are technically feasible intermediate steps to this goal. A 250 MeV linac feeding the present synchrotron would serve a triple purpose. It would replace the existing and technologically old 70 MeV injector, it would increase the number of neutrons by a factor two and would serve as the initial stage of the final 800 MeV injector. There would be advantage, whatever the stage of the accelerator development, in having two target stations to provide the flexibility, optimisation and increased capability as described below.
Although ISIS was designed as a neutron source, its development as a proton accelerator is certain to spawn new facilities. This has been demonstrated by the addition of a muon source and by the choice of ISIS by KfK Karlsruhe for their neutrino facility. There are some ideas, which are at a very early stage, such as the production and study of very short-lived radioactive atoms. -These novel uses will be explored. Within the next few years it is hoped that the ISIS muon facilities can be developed within the international context through the recent proposal to the European Community Large Facilities programme. This would provide three experimental stations with improved quality in place of the present one, particularly for muon spin resonance experiments. There is also a plan, in collaboration with the Japanese (who now expect to have funding approved), to install a new high performance muon line from the same muon production target, to extend basic studies of muon-catalysed fusion.
All the developments described above will require matching facilities at RAL for the user community. These are scientific, involving provision for preparation and scientific categorisation of experimental samples, and organisational support. The instrument scientists at ISIS must be able to interact effectively with the user community and to have the vision to develop the scientific potential of the facility. This requires that they must have their own properly·supported research programmes. It is planned to reach the level of 10% of the direct ISIS manpower devoted to in-house research. Similarly, if there is to be a development of the accelerator and target systems to the long-term goal, there must be R&D programmes in accelerator and target technology, eg in superconducting radiofrequency devices. It will not be possible to realise these longer term development plans, and maintain operations, without some additional capital investment in the mid-to-late 1990s.
The facility described for the 1990s would be of such a class that it should clearly be considered in a European context. The Neutron Review Panel is likely to consider this aspect. The Laboratory plan would be to continue discussion to encourage an extension of the international contributions beyond the present 17% level either by extending the present bilateral agreements or through some full internationalisation programme. There would be merit in jointly optimising the facilities at ISIS and the steady state reactor at ILL to give UK (and other European) scientists the widest range of neutron-facilities for condensed matter research which are unlikely to be surpassed into the next century.
Established in 1976, the Central Laser Facility (CLF) has a record of successful operation, a diversifying scientific programme, increasing user numbers and developing facilities over the last decade culminating in current provision of a 5 TWatt Nd glass laser VULCAN, a prototype high power KrF laser SPRITE and the Laser Support Facility (LSF), comprising a range of smaller picosecond and frequency-tunable lasers in both a loan pool and in CLF based laboratories.
The most recent review of the CLF in 1988 by a Panel chaired by Professor Challis, confirmed the important role of the CLF in supporting the multidisciplinary research of some 125 academic staff in HEIs and a similar number of students and RAs. The high power Nd glass laser VULCAN would serve the needs of the high power laser users for a few years with the construction of an advanced new KrF laser SUPERSPRITE to supersede VULCAN in 5 to 6 years. SUPERSPRITE would out-perform VULCAN by up to 100 times in peak power of UV picosecond pulses and seven times in UV energy for approximately the same capital cost, giving users a facility that would be at the forefront of the development of new laser technology. The Review Panel also recommended a modest increase in the scale of the LSF, recognising its cost effectiveness, large user demand and scientific productivity.
The longer term future of the CLF may be significantly influenced by current interest in a European High Performance Laser Facility which has been identified in SERC's Corporate Plan and is the subject of study of a Scientific and Technical Working Group established in February 1989 by the research agencies of France, Germany, Italy, Spain and the UK.
Science Board, in approving an initial R&D phase of work on SUPERSPRITE, decided to have a mid-term review of the programme in the light of international developments. The scale under discussion for the European facility is a factor of 30 greater than SUPERSPRITE and comparable with the largest lasers in the world. Experiments using a laser would involve relatively large teams with substantial preparative effort. The end product would be a world-leading programme of research. Participation in this programme by UK HEIs would require coordination, preparation of experiments and training of students on a smaller scale domestic facility. The nature of such a supporting facility would depend on several as yet indeterminate factors including the location of a European facility and the choice of technology (KrF, Nd glass or hybrid). The adjustment of SERC's programme in relation to developments in Europe and the exploration of the possibility that RAL could be the site for a European laser and KrF the choice of technology will be important tasks for the next few years.
The need for access to small lasers will remain uninfluenced by any European High Performance Laser Facility. Lasers are evolving rapidly and the use of both presently available and new devices in multidisciplinary research will expand and stimulate further demand from users for centralised access to state-of-the-art systems. SERC will wish to service this need; the Challis Panel concluded that the task was best continued at RAL where co-location with the high-power lasers gave access to a pool of laser expertise which was not available anywhere else in the UK. They recommended that this should be reviewed in three years' time to take account of the developing needs of the research community. The European developments will also clearly influence the argument over this period.
The need for services of the type provided by the CLF will continue into the 1990s but the pattern of provision will change if SERC becomes a partner in a European High Performance Laser Facility. RAL will make every effort to exploit the novel KrF technology and to secure a key role in a European facility.
RAL's activities in Information Technology in the 1990s will be significantly different from the activities in the 1980s. There will be less need to focus on standard hardware and software platforms for the academic community. The move towards distributed processing has resulted in a degree of standardisation within the industry. The increase in communication bandwidth over the next decade will allow the rise of specialised server facilities remote from a particular site. RAL is ideally placed to provide server facilities such as specialised software, databases, storage, computer power etc. Information Technology is both a major activity within SERC in its own right and an enabling discipline relevant to many other areas.
The computing requirements of the academic research community and industry will continue to rise by a factor of two per year throughout the 1990s. The likely inability to provide all such general facilities via central scalar or vector computers past 1994 will place a greater emphasis on parallel computing and distributed supercomputing. The early 1990s will see an explosion of different architectures, languages and techniques before de facto standards appear. RAL is already involved in a number of activities relevant to this area - the Transputer Initiative, assessment of the new multiprocessor superworkstations, quality software development, interactive design and computer graphics. RAL's role will be to participate in the exploratory developments and provide infrastructure as required.
The continuing advances in workstation capabilities will mean that by 1995 substantial engineering design activities incorporating computer modelling and simulation will be possible interactively. This will have a significant effect on the design process itself and will be a major challenge to the engineering researcher. RAL's role is to ensure that mature IT advances are conveniently accessible by the engineering researcher as early as possible.
It is expected that the progress of advanced micro-electronics and integrated optics devices will depend on HEI research workers having access to high precision micro-fabrication facilities that, at the leading edge, can best be provided centrally. RAL's Technology Department will continue to operate advanced facilities (including that supported by the National Electronics Research Initiative) for electron beam lithography (EBL) and its successor techniques, with increasing emphasis on nano-technology. A suitable programme balance will be maintained between support to HEIs, R & D for UK industry and participation in European initiatives. The current role of the RAL EBL Facility as an academic-industrial research centre for advanced lithography will develop a European dimension. By the late 1990s the EBLF could become a major European academic-industrial centre for advanced microstructures.
Electronics Division at RAL provides an "end to end" design and fabrication service for SERC grant holders in more than fifty departments at 38 HEIs. All Boards of SERC use the facilities, which provide access to a comprehensive range of software and hardware for- implementing electronic· systems using silicon, GaAs, hybrid or circuit board technology. The success of the UK activities has led to an increasing European role which will become a major activity in the 1990s. The Division will provide the European academic research community with the best available electronic technology, and hence enhance both scientific and engineering research.
As a result of increasing world-wide interest in renewable energy sources and 'clean' generation of electrical power, there is a strong and growing wind energy research community in the UK and in Europe. RAL has responded to UK university needs by setting up a wind energy test facility, coordinating its use, and providing collaborative support for the associated research. Some ongoing projects also involve participation of UK industry. RAL will continue coordination and facility management in these areas. Additionally, a modest amount of energy related repayment work will include projects with the European academic community (eg through EC programmes) and with European industry. It is not clear how far into the 90's support for wind energy research (as distinct from exploitation of the technology) will continue. Some work on novel concepts and exploring new potential applications will probably be maintained, but it is difficult to judge the level. On the other hand, RAL is well placed to provide support to the development of wider interests in research on environmentally-attractive technologies, either through the provision of medium-scale facilities, coordination or both. This area of activity, although fairly small by laboratory standards, has scope for making a continuing contribution to Engineering Board's objectives.
RAL will continue to provide facilities and services that are needed by the Engineering Board. These will be medium-scale rather than large-scale, but this does have the advantage that the Laboratory will have the flexibility to respond to changes. It is particularly difficult to forecast what will be needed in detail in the support of information technology; for example, ten years ago the impact of the transputer or of parallel processing on current programmes would not have been predicted. What is certain is that technical support in information technology, electronics and engineering will be required by the research community for some time to come. Engineering Board will also benefit from the continuity of expertise at RAL in providing consultancy and assistance with setting up programmes; this role is expected to become increasingly important in areas such as software, networks and engineering design.
The main thrust of changes to the Laboratory's infrastructure in the 1990s will be to maintain a steady improvement in administrative efficiency that has been a feature of the development of the Laboratory over the last five years, and to provide buildings and services suited to the role of the Laboratory to the end of the century.
There is an urgent need to replace the last major temporary wooden building on the site, building R20. Provisional plans have been drawn up for a replacement building that will co-locate the principal elements of the Laboratory's Administration Department including Finance and Accounts, Personnel, General and Scientific Administration including Public Relations. The new building is estimated to cost between £ 2M and £ 2.5M. Under the new RMS arrangements the funding for this building could be found from the proceeds of sale of capital assets including Chilton House, further Laboratory houses and, possibly, the remaining parcel of land in the parish of Chilton.
If the Laboratory becomes the site for a Eurolaser project, new buildings will be required. It is proposed to plan on allocating the former Proton-Linac complex, Building R12, with suitable additional buildings for this purpose.
The steerable radar dish at Chilbolton will reach the end of its useful life before the end of the 90s. When this time is reached the Chilbolton outstation will be disposed of.
RAL depends on Harwell for many services including fire, medical, heating and transport. As Harwell moves further into a competitive industrial regime it may be that these services will become more expensive. It is believed that alternatives can be found including the reliance on public fire and medical services and installing an 'on-site' steam boiler to provide steam heating services. A major problem for the Laboratory remains that of transport. The Laboratory is not viable in its present location without official transport provision; some 700 people rely each day on the Harwell bus service to get to work and it is difficult to envisage a regime in which RAL would not depend on Harwell transport.
The present 'home made' management accounting system is no longer adequate to provide the service required effectively to manage the Laboratory. It is anticipated that the early 1990s will see the full implementation of the MSA package; this will require substantial re-organisation in the financial management and reporting procedures in the Laboratory.
The last two years have seen a major effort to plan for the introduction of formal training at all levels in the Laboratory. In the 1990s it is expected that each member of staff will spend 2-4 days each year on formal training. It will require special care, and will not be straightforward, to ensure that adequate funding is provided for training and that training schedules can be arranged for staff who are under pressure from project needs.
There will be sustained emphasis on publicity and public relations activities and it is expected that the public relations group in the General Administration Division will be enhanced by one post to deal with the increased load. Consideration will be given to following the lead of the observatories in establishing a Visitor Centre on the site as part of the replacement for building R20.
The new departmental structure gives the possibility of a more effective distributed departmental administration structure. It is planned that in each Department, departmental resource managers will command small departmental administration offices to provide front line support to the scientific programme departments.
Plans for repayment work are dealt with in Section 8. The increase in activity on repayment work has infrastructure implications and these may lead, in the 1990s, to the need for a commercial office as a separate entity within Finance and Accounts Department. The commercial office would deal with contracts, estimating, tendering- and monitor cash flow and profitability in the Laboratory's commercial activities.
Federal Computing takes about a quarter of the overall effort used in the Establishments on computing. It covers large scale central (super)computing with the networking and other infrastructure needed to support it. For the future it may be desirable for there to be some federal support of activities in novel architecture computing as a means of improving coordination and effectiveness.
The 1989 SERC Corporate Plan includes: 'Council wishes to see the Atlas Centre continue to develop as one of the world's leading supercomputer centres'. The Centre already houses, as a joint Research Councils' facility, the UK academic community's most powerful supercomputing complex and it will play a leading role in the national provision as it develops through the 1990s. The Centre's joint Research Councils remit could be extended into other areas of large scale computing in interdisciplinary areas or where there would be economies of scale.
State-of-the-art computational science projects in the 1990s will require one or two orders of magnitude more computational resource than can be provided on today's supercomputers. The number of such highly demanding projects from all Research Councils will be relatively small (tens rather than hundreds) and will be feasible only on the most powerful computational facilities obtainable. Because of their cost such facilities, whatever their nature, will be central. They can be expected to have ten times the power of today's Cray X-MP/48 by 1992 and a hundred times by 1997 at roughly constant price.
During the 1990s affordable departmental and desktop machines will attain the power of today's supercomputer; the trend is clear, if not the timing. Such machines may well be adequate for all but the most demanding problems. They will not themselves be a call on federal funding but they will need access to networking and other federally supported services.
It is the role of the Joint Policy Committee for Advanced Research Computing to propose to ABRC and others the nature and scale of future national supercomputing provision. A proposal is being prepared now for a next generation supercomputer (inter alia) for about 1992, and a further generation of provision would be timely in the late 90s though it is premature to speculate on its precise nature.
Over the next few years there will be a substantial growth in power of both multiprocessing conventional architecture supercomputers and of novel architecture ensembles. Trends suggest that there may be little difference in the ultimate power attainable from conventional and novel parallel machines at the same price level, although the highly competitive nature of the business makes this a dangerous statement and there will always be claims made for disputing it. It remains to be seen whether the novel architectures can be made easy to use for general purpose computation. At the level of the discussion here the important point is that some form of massive supercomputer will be required for the most demanding research.
There is a documented need for continued conventional mainframe computing at RAL into the 1990s for particle physics data reduction from LEP and HERA experiments. ln the later 1990s other options for handling this type of work may become viable and these will be investigated in collaboration with users.
As well as supplying processing power, major roles for the Atlas Centre will be the provision of advanced facilities for the storage, retrieval and archival of the vast quantities of data associated with state of the art projects, and the provision and support of very high performance and functionality networking.
The skills required at the Centre will need to evolve. It will be vital to sustain and develop, expertise in fast communications, visualisation, parallelism and algorithms since these will be key to the productive use of powerful equipment. Operator intensive work will reduce.
The Corporate Plan states that: 'The optimum ways of utilising novel architectures in Science and Engineering will require expanding support'. The potential of novel architectures needs to be explored in depth, in ways closely coupled to scientific problems, and across a variety of machine types. The main issue is that the ability to develop hardware ensembles is currently much more advanced than the ability to produce software that allows this hardware to be fully and easily exploited.
Daresbury Laboratory is already pursuing particular applications currently based on the Science Board programme. It has skilled software people and a wide science base resulting from SRS and CCP support activities and is working towards an effective and portable software environment. More effort is required to speed the development and to support its exploitation, and there may be a need for powerful hardware to act as a focus and to handle work greater than can be handled on small dedicated systems.
The Engineering Board's EASE programme at RAL and the Transputer Initiative contain activities in novel architectures, as increasingly will other Board programmes at all laboratories.
Peer Review via Boards couples the existing work in laboratories with complementary activities in HEIs and must be preserved. It also couples the activities to real user needs. A federal dimension to this work is needed to give the most cost effective use of resources, rapid dissemination of expertise through the community and to promote cross-Board interaction. Also, the technology could become pervasive: novel architectures have potential not only for supercomputing, but also for lower levels of computing and for embedded instrumentation.
It is expected that there will continue to be a need for large central supercomputers for the highest power applications and it is appropriate to develop the Atlas Centre at RAL to meet this need, both within the Research Councils and more widely. At the same time, while desktop computers may provide sufficient power for most applications, there will be a continuing need for central support activities such as infrastructure and networking. The potential applications of novel architecture computing should be explored through a wide-ranging programme building on existing strengths and expertise at RAL and DL as well as in HEIs.
The laboratories have done a tremendous job over the last 10-15 years in equipping the UK research community with world-class scientific facilities. These are now proving their worth through their output and this will continue if the policies in the 1989 Corporate Plan can be sustained. There are a number of good options for the next generation of capital developments to take the UK, working with other partners (especially in Europe), into the 21st century with high-class facilities. The laboratories will work with HEIs to realise these opportunities, making the necessary adjustments to the balance of their programmes so that resources can be channelled into new projects as older activities are displaced. The broad features of these developments must now be translated, through the usual processes, into detailed plans, as is already happening in some cases.
