Minutes of the meeting held on 6/ 5/1964 at 11, Charles II St, London

• Present
  • Sir William Penney (Chair)
  • Dr J B Adams
  • Professor B H Flowers
  • Dr J Howlett
  • Mr C Jolliffe
  • Professor T Kilburn
  • Mr H J Millen
  • Professor R E Peierls
  • Dr T G Pickavance
  • Sir Graham Sutton
• Alternates Present
  • Dr S Hollingdale representing Mr M J Lighthill
  • Dr W Marshall representing Dr F A Vick
• Secretary
• Apologies
  • Dr R A Buckingham
  • Sir John Cockcroft
  • Dr J C Kendrew
  • Sir Harrie Massey

Sir William Penney said that this was the last meeting in which he would be taking part. He welcomed to the meeting Professor Flowers whom Lord Bridges had invited to succeed him as Chairman. Sir William Penney also introduced Mr H J Millen, a finance officer of the AEA at present attached to NIRNS who had been invited to be a member of the Committee.

1 MINUTES OF THE LAST MEETING

The Committee approved the minutes of the meeting held on 13th November, 1963.

2 MATTERS ARISING FROM THE MINUTES

Minute 3.1: (Adequacy of the main output system) Dr Howlett said that he had looked into the point raised at the last meeting. It was of course true that in certain circumstances the output system could be overloaded but its capacity was large (8,000 words per second) and the system also provided for smoothing out peaks. in his judgement there was no reason at present to believe that the design was inadequate.

Minute 3.7: (The Atlas Laboratory programme of work) Dr Howlett said that he had not yet prepared the programme as it would on policy decisions on several points which were at present under discussion. the Committee agreed with Dr Howlett in not being unduly hurried in preparing the programme but the view was expressed that it should be prepared and discussed before plans were made for requesting any further substantial capital expenditure.

Minute 5: (Estimates 1964/65) It was reported that in preparing the Estimates 1964/65 for final discussion with the Treasury, certain changes had been made from the figures in paper NI/ACC/63/8, namely

1. changes in the incidence of capital expenditure between 1964/65 and other years, arising from later knowledge of the progress of contracts;
2. provision for additional non-capital expenditure corresponding to a complement of 60 by the end of the year.

3 PROGRESS REPORT BY DR HOWLETT - NI/ACC/64/1

Arising out of Dr Howlett's Progress report the following items were discussed in particular:

3.1 Computing time: Tests at the maker's works indicated that computing times were within those specified with the exception of the time for division which ranged from 10 - 30 microsecs with an average of 25.6 compared with a specified average of 17.9 microsecs. Professor Kilburn said that this was due to a recognised logical error in the divider which he had decided not to rectify because the time taken to do so would have been considerable whereas the effect on the overall speed of the machine was small. the overall speed originally expected had been 60 - 80 times that of Mercury and the overall speed now found was 70 - 80 times that of Mercury.

It was noted that the formal tests of the machine would take place after installation in the Laboratory and a decision whether to take any action as a result of low dividing speed would then be needed. in the meantime Dr Howlett was asked to consult the contracts department as to whether any representation should be made to the contractor at this stage.

3.2 Drums: Dr Howlett said that there had been two sources of trouble with the drums. Firstly the cross talk which had been overcome and secondly mechanical troubles of accurate construction. The latter was not yet wholly overcome in the drums of British manufacture and the contractor had arranged to obtain American drums as an insurance. Whether these were finally used or not there would be no additional charge.

3.3 Data links: With reference to the newly available equipment for rapid transmission over ordinary telephone lines, Dr Howlett made it clear that in his view users should be responsible for the cost of the equipment at their end. the Committee considered that equipment of this type was likely to be of great value and asked Dr Howlett to keep closely in touch with developments. the use of high speed transmission lines as used for television seemed on present information likely to be a very expensive matter and probably not advantageous.

In connection with this matter Dr Howlett asked for advice on the allocation of fixed times to individual users for data transmission because this might come to represent substantial amounts of priority time on the computer. it was agreed that this was a complicated matter because of customers demands for a very quick turnover. Nevertheless the questions of transmission time bookings and machine time bookings must as far as possible be disassociated. Dr Howlett was asked to report further on the problem when he had more detail and in the meantime not to commit himself to individual customers.

3.4 New IBM peripheral equipment: Dr Howlett was asked whether any of the new IBM 360 series equipment could be used to provide much improved facilities on the Atlas. he said that he would keep in touch with development of this as of other new equipment but at present the Atlas installation appeared to have a well planned and satisfactory range of peripheral equipment.

4 MINUTES OF THE ATLAS USERS COMMITTEE

The Committee took note of the minutes of the second and third meetings of the Atlas User Committee. in connection with the first paragraph of minute 2 of the third meeting Dr Howlett was asked for information about the scale of the work mentioned as being sponsored by the Atlas Laboratory on the London Atlas. He said that this was quite small, the total cost being a few hundred pounds.

5 FIVE-YEAR FORECAST - NI/ACC/64/3

5.1 Staff: In considering the five-year forecast submitted by Dr Howlett the Committee first dealt with the question of staff numbers. There was a general discussion in which the following main points were agreed:

1. Operating staff: It was agreed that the necessary number of operating staff should be provided but that the numbers which would be required were not yet clear. Figures as proposed in the paper should go into the five-year forecast and would be progressively reviewed in later years.
2. Build-up of shift operation: It was agreed that the build up to four-shift operation should be as rapid as possible as proposed by Dr Howlett. On the other hand it was considered that the rate shown for building up to four shift operation was optimistic and should be relaxed by a few months.
3. Research staff; also expenditure on extra mural research: While confirming the importance of an element of mathematical research by the Laboratory's own staff, the Committee considered that the number of such research posts should be shown as not rising above seven in the five-year forecast and that the number of programmers directly supporting them should be correspondingly limited. The hope was expressed that the number of research mathematicians working in the Laboratory would be supplemented by some university mathematicians.
   Dr Howlett was asked about the nature of the extra mural research shown in paragraph 3 of the paper as part of the cost of research activities. He gave as an example the writing of certain compiler programmes in universities and the point was made that part of the sum might better be described as operational expenditure than research expenditure. It was agreed, however, that the amount shown in the five-year forecast (Table D) should not rise above £30,000.
4. Complement 1964/65: The Committee approved a complement of 60 for the Atlas Laboratory for the current year 1964/65.
5. The following further points were also made in the discussion:
   1. Professor Flowers said that he would like to discuss the staffing as well as the financial proposals in the five-year forecast in detail with Dr Howlett. After consideration it was agreed that any changes resulting from these discussions would probably have effect soon enough if they were incorporated in the next five-year forecast.
   2. It was suggested that the distribution of staff numbers between programming and operating might have a substantial effect on the efficiency of the Laboratory. this would be one of the matters to be discussed between Professor Flowers and Dr Howlett, who would of course welcome detailed discussion with any other member of the Committee.
   3. The proposed staff numbers were close to those used on the Aldermaston Stretch. There was some discussion on the difference between the two installations and it appeared that the major difference lay in the wider range of users of the Atlas.
   4. It was also stated that the forecast annual expenditure was about twice that of the Manchester or London Atlas Laboratories, and Dr Howlett was asked to look into this. The Committee realised that the difference might arise solely from accountancy procedures but they wanted to be clear.
   5. The arrangement already made that university problems for the Atlas would normally come through the University Computing Laboratory would be most valuable in increasing efficiency by diverting problems more suitable to smaller computers.

5.2 Allocation of time:

1. The Committee discussed the question of allocation of time in view of the indication in the five-year forecast that demands would very soon exceed the time available. It was recalled that the Committee had decided after detailed discussion at the meeting on 27th November, 1962 that they themselves would deal with major principles of allocation, perhaps setting up a Priorities Sub-Committee to help them. It was agreed to wait until there was actual experience of pressure of work before dealing with this problem. However, three suggestions from practical experience were made by Members:
   1. that the allocation of limited amounts of time to users in advance was a powerful incentive to them to write efficient programmes;
   2. that any judgement of the worthiness of jobs was best done after the event using referees where necessary and the results applied to future allocations of time;
   3. that in some cases jobs would be accepted at low priority with the warning that they would be delayed.
2. Nuclear Physics: Attention was drawn to the limitation of time allocated to nuclear physics to 40 hours per week compared with the estimated demand of 100 hours.
3. Medical Research Council: Dr Howlett said that he had not as yet approached other units of the Medical Research Council besides Dr Kendrew's Unit with regard to their possible requirements and enquired whether he should do so. the Committee asked him to write to Sir Harold Himsworth.

6 ESTIMATES

The Committee considered the draft Estimates for 1965/66 presented in paper NI/ACC/64/4 and in particular they questioned the provision of a substantial sum for building extensions in the estimates year. It was decided to postpone the provision for one year. With this amendment and subject to any further discussion at a later meeting the Committee approved the draft Estimates.

7 FUTURE MEETINGS

It was agreed to hold future meetings at the Atlas Laboratory and the Secretary was asked to try to arrange the next meeting either in the first week in September or early in October. It was agreed that the future work of the Committee should be reviewed at the next meeting, the suggestion being made that perhaps the Committee should include the reviewing of computer requirements and developments on a much broader scale. The Committee would also clearly have to consider the Estimates and the five-year forecast for the Laboratory.

The frequency and timing of future meetings should also be considered at the next meeting and the point was made that it would be convenient to hold one meeting in the spring to consider the five-year forecast as well as technical matters and one meeting in September to consider the Estimates as well as technical matters and perhaps a third meeting to consider technical matters only.