Minutes of the meeting held on 26/ 4/1968 at State House, London

• Present
  • Lord Halsbury (Chair)
  • Dr R A Buckingham
  • Professor T Kilburn
  • Professor E S Page
  • Professor R E Peierls
  • Dr T G Pickavance
  • Mr L F Rutterford
  • Professor M V Wilkes
  • Dr J Howlett
  • Mr C Jolliffe
  • Dr J H Price
  • Mr R St J Walker
• Apologies
  • Dr A H Halsey

1 Minutes of the Last Meeting

The Committee approved the Minutes of the meeting held on 17th November 1967 with the following amendments:

• Item 4, first paragraph: delete the words perhaps a little lower
• Item 7(6) on page 4: the last phrase to be altered to read partly to let the Atlas Computer Laboratory put more of its effort into software development, from which the British industry would benefit

2 Matters Arising

Users' Committee: Dr Howlett told the Committee that he hoped to arrange a meeting of the reconstituted Users' Committee in late May or early June.

3 Progress Report - ACC/68/1

The Committee noted the various tables giving the numerical facts about the work-load and the machine's performance. The following points were made:

1. The throughput of the installation continued to increase, due partly to larger working hours and partly to increased efficiency. Dr Howlett said that the engineers had cut the daily maintenance from 4 hours to 2, and that 22 hours was scheduled for computing service each day; the working week was now extended by 8 hours to Saturday morning, and would be further extended to Saturday midnight by the Autumn. The performance record would be studied critically at the end of May, to see how the machine had stood up to the reduced maintenance. The engineers had said that they would always want possession of the machine for a full 8-hour shift on Sunday.
2. The routines for allowing library programs to work from the disc, and private storage needs to be allocated, were in use.
3. The hardware needed for the modification of the Teletypes was to be delivered on Monday, 29th April
4. The demands from Harwell and the Rutherford Laboratory having fallen to small values, there was little reason to continue to report these as separate groups. Harwell usage might in future be included in the External (i.e. paying) group and Rutherford in the Internal (non-paying group).
5. The increased use by universities was striking (4,608 hours for the 1967/68 financial year, against 2,742 for 1966/67), especially the very rapid increase by the smaller and the new universities.
   (N.B. These times, given in table 5, were not clock times, but instruction interrupts converted to time by the relation 1 interrupt = 12 milliseconds; in present circumstances this over-estimates the clock time by a variable amount which can go up to 20%).
   It would be interesting to see how these demands changed in the future as universities got their new computers into operation.
   The Committee noted requests for better documentation of the Supervisor and for completion of the LISP compiler.

4 Estimates 1968/69 and Forward look 1969/70-1973/74 - ACC/68/2

As no action was needed, the Committee noted this paper: Mr Walker said that there would almost certainly have to be some changes, for example in numbers of staff, in response to further restrictions imposed by the treasury on the SRC as part of the Government's general policy of economy in public spending.

5 SC4020 Microfilm Plotter: Costs and Charges - ACC/68/3

The Committee accepted the paper and authorised Dr Howlett to raise charges for the use of the SC 4020 at the rates proposed; and asked him to review the outcome towards the end of the 1968/69 financial year.

6 New Computer for the Atlas Laboratory: ICT Revised Project - ACC/68/4

In discussing this paper the Committee were very sharply aware of two disadvantages:

1. ICT had still not produced a specification of the P.51 system.
2. The recent merger of ICT, EEC and Plessey had disturbed the whole project, and the effect of this disturbance was so far unknown.

The following points were made:

1. The change from the original 3-processor system, extendable to 4, to one of a maximum of 2 processors removed some of the flexibility but simplified the software; it was probably inspired partly by ICT's fears of software complications, partly by their interest in producing a twin-1907 system.
2. There should be no difficulty in producing hardware capable of 7-10 times Atlas speed, but the production of software to run even this simpler system efficiently would be a serious problem. It would be worth urging ICT to deliver the hardware as quickly as possible, and accepting the need to develop software on the site, in collaboration with other customers.
3. One particular weakness of ICT's proposal for software was the holding of a complete operating system in each processor; this was made necessary by a decision to allow peripherals to be attached directly to either processor. A simpler and more efficient system could be designed if all peripherals were attached to one, the other communicating only with the disc and the main core store.
4. The Manchester Regional Centre were considering a system similar to that now proposed for Chilton. They too considered the mass core store essential for operational efficiency.
5. There was a great need for a serious study of the problems of linking processors, and by tackling the problem with a full-scale installation the Chilton laboratory would be making a valuable contribution to the technology of computer software.

Asked about his own work on a new design of computer, Professor Kilburn said that it was a research project and that at this stage he could not advise the Committee to base any plans on its outcome.

Professor Page emphasized that the development of the Laboratory was being held up because no firm information about the production of the P.51 could be obtained. The 1906A, however, was certainly being produced and advantage could be taken of this fact; he suggested the Committee should consider proceeding as follows:

1. order the 1906A, with peripherals, disc and mass store as in the paper, to be got into operation as a single processor as quickly as possible;
2. order a second 1906A central processor, either on hire or with an undertaking from ICT to take this back on acceptable terms when the P.51 processor was produced;
3. start to attack the software problems by linking the two 1906A processors.

The Committee recognised that in the present state of uncertainty of the P.51 project they could not make a firm recommendation to the Council, and expressed their gravest concerns lest the delays and uncertainties which had operated over the past two years should continue to frustrate the development of a large British computer. Dr Howlett read a letter he had had from Mr P D Hall, marketing Director of ICT, dated 22nd April 1968. Mr Hall said that his company was engaged in an intensive study of the effects of the merger on their technical, production and marketing positions, and hoped to present firm proposals within a few weeks of the date of his letter. The Committee agreed to meet again at 11.00am on 4th June 1968, provided that ICT had by that time made their plans known.

7 Consultation with UST Board Committees - ACC/68/5

The Committee commented that, additionally to its prime object of getting an expert opinion on the scientific value of the work, the proposed policy of consulting the USTB committee would have the valuable effect of keeping the Board informed of the nature and extent of the Laboratory's contribution to university research. They requested Dr Howlett to proceed along the lines of the paper, suggesting that he should agree the details of the procedure with Mr Jolliffe.

8 Distribution of Atlas Laboratory Programs by the National Computing Centre - ACC/68/6

The Committee agreed that it would be to everyone's advantage to have the Laboratory's programs disseminated as widely as possible, provided that the universities' position was not prejudiced and that the charges made to industrial or commercial users were realistic and equitable. They approved Dr Howlett's proposal to open formal discussions with the National Computing Centre, involving the SRC Finance Branch.

9 Date of Next Meeting

See last paragraph of item 6.