1. The Minutes of the seventh meeting were approved.
2. Dr Howlett reported that the FR80 had been delivered to the Laboratory on 19 March 1975. It was hoped to turn the power on while the meeting was in progress. He said that his remarks about the commencement of a service in April were optimistic, but the machine would be brought into service, probably in May.
3. Dr Thomas commented on the provision of dial up data transmission at 2400 baud. He said that the Post Office had certainly improved the position at 1200 baud and had not given up their attempts to provide a service at 2400 baud. Because of this he felt that the move to use Codex modems might be premature and lead to unnecessary incompatibility with many universities.
4. Professor McWeeny said that his experience was that many users would be content with transmission at lower speeds. He felt that the emphasis on high speed transmission was perhaps ill advised. Dr Howlett commented that the provision of a high speed data communication service had come about in response to the needs of users. Mr Fossey said that the effective rate of transmission on high speed lines could often be improved if transmission in either direction were interleaved. On the ICL 1906A interleaving was not yet possible whereas on the IBM 370/195 with HASP it was. This was a material factor in the choice of line speed for use with a remote job entry workstation.
5. The Chairman said that there had already been substantial discussion within the Committee on this subject. More recent developments had changed the pattern of proposals previously discussed. He invited Professor S F Edwards, the Chairman of SRC, to introduce the subject.
6. Professor Edwards said that he was glad to meet the Committee and discuss with them the regrouping proposals which had arisen at a very important stage in the development of SRC's computing activities. The Council had recognised that the provision of computing facilities for its activities needed greater coordination, especially from a funding standpoint The cuts in SRC funds had put more emphasis on this need. The Council wished to support, within its budget, major new developments in science and could only accommodate the consequent increased expenditure by corresponding reductions in funding for its other activities. In reality the Council faced a severe cut in its liquidity.
7. Turning to the computing scene, Professor Edwards said that the Council recognised that in Great Britain there was a very substantial provision of computing power within the university field, particularly by comparison with that in other parts of the world, even in the United States. There was considerable variability in the way in which its use was organised. In the present financial climate, the Council had to think about the continuing need for its own three major computing centres. The computer installation at the Rutherford Laboratory provided support in the main for the high energy physics experimental programme but also for other areas of science. The nuclear physics community had indicated with reason that a change in the management of this installation was unacceptable. The computer arrangements at Daresbury, although a smaller version of the Rutherford Laboratory providing a very good service, were faced with a completely new set of circumstances: the decision by Council to close the support of the high energy physics programme at Daresbury in order to save funds; the installation and use of the new Nuclear Structure Facility; and a major event in the traditional disciplines of the Science Board, the Synchrotron Radiation Source. The latter would give the Science Board access to two ma1n sources of radiation, X-rays at Daresbury and neutrons at ILL and Harwell. These facilities would represent a major contribution to science within the UK comparable in scale to those provided by SRC for high energy physics. It was likely that much conventional X-ray crystallography would be superseded by experimental techniques based upon the new facilities. Theoretical study at Daresbury would become an important component of its activity, the more so as the new facilities come increasingly into use. Against this background arose the proposal to move the Atlas Laboratory to Daresbury. It assumed that the support for research in the computer science area would remain largely in the universities.
8. Professor Edwards said that the picture had changed following the study carried out by the Engineering Board into computing needs in engineering research. The report had stressed the need for centralised interactive computing facilities. Their provision could have been the responsibility of any of the three main SRC computing outfits. However the Department of Industry had proposed a national federal computing institute at Chilton, since it had found a need for a stronger computational base in Britain, the existing resources both in the Department' and outside being somewhat fragmented.
9. Professor Edwards said that SRC was faced with a situation comprising the maintenance of the Rutherford Laboratory as the main source of support for high energy physics, the development of Daresbury into a major laboratory with large scale facilities for experimental work in science, the demand in the engineering community for an interactive service, and a request from the Department of Industry to play a full part in a new federal computing institute. The Council had therefore considered a solution in which the present Atlas Laboratory buildings would become part of that institute with some staff and possibly others to run the interactive service from Chilton. This arrangement would match the requirements of the Department of Industry and its part in the new institute.
10. Professor Edwards said that the remaining part of the Atlas Laboratory, especially that already identified with the existing service to the Science Board, should logically move to Daresbury where it would become part of an enhanced and more important computational facility on that site.
11. Professor Edwards said that with these arrangements SRC would continue to provide a service for high energy physics at Rutherford Laboratory, participate in the national federal computing institute, and integrate the Science Board activity on the Daresbury site. The Council would be able to reorganise its activities within the cuts already imposed by the Government. Important consequences would be that funds could be available for the purchase of equipment for the new engineering interactive service, for the provision of a new large computer to meet the needs of computational science, and thirdly that funds for new equipment in the federal institute might be provided in unison with the Department of Industry. Overall the Council would be able to take advantage of its substantial investment in new large scale scientific equipment and to provide the computing disposition to exploit this equipment. It could also continue to provide the support for projects depending on large amounts or computing time.
12. The Chairman invited the Committee to discuss the regrouping proposals by treating firstly, the concepts of dividing SRC computing outlined by Professor Edwards into five broad areas; the support of high energy physics concentrated at the Rutherford Laboratory, the development of new science facilities at Daresbury, the participation in the federal institute at Chilton, the Engineering Board's centralised interactive computing facility, and the provision of a service in large scale computing to a heterogeneous set of customers: secondly, the question of the location: and lastly the matter of the management of the facilities.
13. Mr Taylor felt that the Department of Industry's position had been well summarised. He emphasized that bringing together the separate strands geographically had important benefits. He stressed that there were many points under discussion between SRC and DoI on which firm statements were not yet possible. The views of the Committee on the principle of the scheme would be of value in subsequent negotiations.
14. Mr Seddon said that NERC was extremely conscious of the pressure for the provision of computing facilities for some of its University grant aided programmes. Like SRC, NERC was concerned that, although the Computer Board had provided adequate and substantial facilities to the Universities, they were not always being used to best advantage for important scientific work. As a result, SRC was being asked to provide the requisite facilities from its own resources. This was a matter for serious consideration in future by both NERC and SRC.
15. Professor Burke said that he agreed that there were distinct areas, but he felt that the problem was much more whether the provision of facilities should be in two rather than three centres. The demand for interactive facilities did not arise solely in the engineering field: it could be valuable in science. Since engineering simulation studies gave rise to large computational demands, the need for such facilities was not exclusively in the physical sciences.
16. Professor Barron said that the distinction between interactive and background working was quite artificial. Projects often had a need for both at different stages. He felt that it would be wrong to set up a barrier between the parts. He was worried lest by transferring all Science Board activity to Daresbury SRC would remove the independence of the Atlas function and encourage the feeling that the outside work was a poor relation to the in-house programme. Professor Kilburn said that there was a great deal of substance in these remarks.
17. Professor Kilburn said that ideally computing should not be divided. Practically where some activity occupied nearly all the available facility it was sensible to treat it separately, as had been done with high energy physics computing. The interactive facility proposed by the Engineering Board would in due course need the support of a large computing facility; its ultimate size would determine whether it should be treated separately.
18. The Chairman said that because the high energy physics community was close-knit there was much wisdom in treating their computing needs separately. Although the NSF and SRS were important instruments of research, the community of users was not confined to a single discipline. He felt that in spite of the significance of SRS there was less risk that its users would become the dominant group on centralised computing facilities.
19. The Committee agreed that the Engineering Board's and the Department of Industry's proposals had close links which should be preserved, that the Rutherford Laboratory computing load was such that facilities outside would be inappropriate, and that there was no sharp distinction between the Engineering Board's interactive computing requirement and the longer term needs in the physical sciences.
20. Turning to the question of location, the Chairman said that the Committee need only consider Daresbury and Chilton as sites. It would be reasonable to assume that the Rutherford Laboratory would continue to support high energy physics, and that the interests of the Department of Industry and the Engineering Board should be kept together. This meant that the Committee should examine three possibilities:
He reminded the Committee that D of I had expressed a strong preference for the choice of Chilton as the site for the federal institute.
21. Dr Thomas said that the Committee could not give an opinion without some idea of the staffing numbers and the level of expenditure involved. Mr Taylor said that the proposals related to the CAD centre at Cambridge and to part of the computing research at NPL. He estimated that around 100 people might be involved and an expenditure of about £O.75M per annum. Mr Davies commented that the combined work of the CAD centre and his division at NPL already accounted for expenditure in excess of £1M per annum. In answer to questions, Mr Taylor said that the services of NEL were unlikely to be affected and that the proposed federal institute would continue to be closely involved with industry but would not be purely a bureau.
22. Professor Edwards said that it seemed likely that the SRS would develop in much the same way as high energy physics by generating large computing needs. To meet these purely by links with Chilton would not satisfy the Daresbury users for long and the pressure for a further computing centre at Daresbury would become irresistible. He emphasized that the considerations concerned the disposition of manpower rather than equipment.
23. Professor Burke agreed that there was a convincing case for good local facilities at Daresbury. He was concerned that a need for a vector processor had been identified which would require a strong group to develop software to exploit its power. He felt that the group would benefit from being alongside those working on the interactive software for the Engineering Board.
24. Professor Kilburn said that the Universities would have no objection to the siting of the Engineering Board facility. They would however question the move of the Atlas function to Daresbury. There would be the feeling that the independence of the new set up would be lost within the general Daresbury framework. Universities were convinced of the continuing need for an independent Atlas function and would wish to see this preserved. Dr Norton said that the Universities were also widely concerned that access to the 20% share of the IBM 370/195 should be maintained.
25. The Committee was not convinced that the Department of Industry had an overriding case for use of the Chilton site for the federal institute. It recognised the proximity to ICL at Bracknell and to Harwell as important factors. It thought that the advantages of incorporating the engineering interactive facilities with those of science at Daresbury should not be overlooked.
26. Professor Dalitz drew attention to the inconsistency between cuts in the support for high energy physics within the UK and an increased investment in the experimental programme at CERN. It was likely that the removal of the IBM 370/165 from HEP support in the North would place additional pressure on the IBM 360/195 at Rutherford at a time when new experimental devices, increased beam intensity and higher energies all pointed to more sophisticated computing for the reduction of experimental data.
27. Dr Thomas said that the question of location could not be completely separate from that of management. The Chairman said that the Committee had always emphasized the need for the actual and visible independence of the control of the Atlas function even if it were located at Daresbury.
28. Professor Edwards said that there would be a close tie at Daresbury between the experimental and computing facilities. He thought that the unit providing the Atlas function would absorb and provide all the computing facilities for the Daresbury site. The Daresbury Laboratory would have a Director who would not necessarily be coupled with the scientific programme. Turning to Chilton, he said that AERE Harwell might move its computing into the federal institute. It seemed likely that the SRC section of this institute would draw on the main facilities, and at the same time be part, of a larger SRC laboratory on the Chilton site.
29. Professor Burke said that the connection with UKAEA could be an overriding consideration for the set up of the federal institute at Chilton. He felt that it was essential that the development of software exploiting new computational techniques should be contained centrally. As the Committee did not have all the facts on which a decision could be based, the possibility must be left open that the work be centred at Chilton.
30. Dr Howlett said that the formation of a national federal institute at Chilton could be of enormous importance. There was already a going concern at the Atlas Laboratory. Moving it to Daresbury would be very disruptive. It would take a long time to develop the relocated Laboratory. He said too that there was great advantage in the use of expertise in widely differing fields. Recently some of the ideas for the transfer of data in quantum chemical packages had been of value in improving the performance of the G-EXEC system of programs for geological database manipulation. He felt that a long term view should be taken by the Committee. There was no obvious reason for removing the work from Chilton. He would be very distressed if the Committee were now to come down in favour of moving the Laboratory to Daresbury.
31. Professor Burke deputising for the Chairman who had left the meeting said that the Committee generally agreed that the formation of one Laboratory would be the best solution in an ideal world. The Committee would wish that the formation of a federal computing institute in the north should be explored with the Department of Industry. The Committee recognised that it might not be possible to achieve this and would therefore endorse the solution put forward in the paper by the Chairman of the Science Research Council. The Committee wished to be sure that the management of the new Atlas function should be, and be seen to be, independent.
32. Professor Burke introduced the report of the panel whose main objective had been to get a definite assessment of demand together with a tentative examination of the hardware possibilities. It was clear that some problems could nor be done on either SRC central facilities or on those provided by the Computer Board. As present usage of the facilities was around 35 hours per week equivalent 360/195 time, there was already little room to meet the increasing demands. New computer projects needing special facilities were clearly coming forward in the fields of statistical mechanics, plasma physics, astrophysics, and quantum chemistry. The installation of the SRS would also encourage substantial demands. The NSF would stimulate use of some new theoretical models and in the field of plasma physics the new laser laboratory could have very important implications on demand for computing.
33. Professor Burke said that in examining hardware the Panel had looked at computers with a minimum processing power of about an IBM 360/195. There were a limited number of possibilities with vector processing. The Panel felt that in five years time there would be a distinct need for a vector processor.
34. Mrs Paton said that the Science Board would welcome an opportunity to consider the paper. She drew attention to the fact that there was no formal limitation on the level of 20% as the share of the 360/195 for use in other than high energy physics.
35. Mr Taylor said that the Department of Industry was interested in the ICL distributed array processor which offered considerable power for vector operations. He noted that the Panel thought that there was a need for a minimum power for conventional computation and asked how serious this was. Professor Burke replied that the Panel had identified a number of problems where the computing power of a 360/195 would be the minimum acceptable. A vector processor was needed for a number of problems particularly in three dimensions.
36. Mr Taylor said that Table 2 of the report showed certain machines whose processing power was only marginally short of an IBM 360/195. He thought that they had been discounted somewhat too easily. Mr Fossey drew attention to the fact that these were really dual processor systems whose performance was related to the kind of workload to which they were applied. Some of the problems the Panel had identified were such that the amount of store required could result in only one processor being used in the solution of the problem with the consequent severe reduction in performance of the system.
37. Professor Kilburn asked whether the amount of vector processing work had been quantified. Professor Burke said that although the Panel had identified problems where vector processing would be used an exact quantification at this time was not possible. However the Panel was convinced of the need for such a facility in the 1980's. Mr Seddon said that he had noted that there was no reference to the needs of other Research Councils. It was clear that there could be increasing demand from the other Research Councils. Professor Burke said that the Panel was aware of the needs outside the domain of the Science Research Council but had concentrated on the computational support required in science.
38. Professor Burke said that the Panel recognised that a great deal of work would be needed to exploit the special facility of a vector processor. There would be a need to put the group working on this where there was considerable expertise in computer science. The Committee endorsed the Panel's report and asked that it be considered by the Science Board.
39. The Committee briefly considered the report of the working group on engineering computing requirements. The Committee accepted this report as a good idea.
40. The Committee looked at the paper on predictions on future computing demand. Mrs Paton said that the Science Board would accept that there was a need to scrutinise more critically the applications from users for computer time. However she felt that the Board would also welcome a contribution from the Laboratory in assessing the efficiency of existing users' programs. Professor Burke commented that his experience as a user showed him that the Laboratory was already undertaking work of that nature.
