Minutes of the meeting held on 7/
7/1961 at Charles House, Regent St, London SW1
-
Present
- Sir William Penney (Chair)
- Dr J B Adams
- Dr R A Buckingham
- Sir John Cockcroft
- Dr J Howlett
- Mr C Jolliffe
- Professor T Kilburn
- 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
- Dr E N Mutch representing Dr M V Wilkes
-
Secretary
-
Apologies
- Sir Harrie Massey
- Dr J C Kendrew
1 The Scope of the Committee's Work
The Chairman opened the meeting by reviewing broadly
some of the things which he suggested the Committee should do, and some
which they should not.
- The Institute has accepted the task of providing and operating
an Atlas Computer and it was the clear task of the Committee to
supervise and control this activity.
- In the meantime, before the NIRNS Atlas was in use, which might
be in early 1964, the AEA has been asked by the Secretary of
the Minister for Science's office, if they could arrange for
some university use of their large computers, e.g. the 7090
at Risley and the Stretch expected at Aldermaston in May 1962.
The Authority has agreed in principle. The arrangements for
implementing this scheme were not complete but the Committee
might be able to help in framing them and perhaps in operating
the scheme.
- He did not regard it as part of the Committee's job to review
university requirements for computers. The University Grants Committee
has recently reviewed requirements for computers
costing up to £0.5M and has recommended the allocation
of five KDF9 machines. A review on a broader scale
has been made six months or so ago by the Minister for Science's
office, and recently the Royal Society has set up a committee
to review university requirements as a whole. Sir William
thought that the report of the Royal Society Committee would be
sent to the Advisory Council on Scientific Policy.
- There was a considerable common membership between the Royal
Society Committee just mentioned and the NIRNS Committee, but
to clarify further the division of responsibility, Sir William suggested
that one joint meeting of the two committees would
be helpful. One special point which might be discussed at such
a joint meeting was the financing of large university computers.
At present, those costing up to £0.5M were being dealt
with by the University Grants Committee, but the Royal Society Committee,
which would presumably initiate recommendations for
larger ones, was not linked with a source of funds.
In discussion, there was no dissent from the above suggestions.
One question raised was whether there should be some further review of
the Authority. It was agreed that this question should be raised at
the proposed joint meeting with the Royal Society Committee.
2 Name and Terms of Reference
It was agreed to call the committee the NIRNS Atlas Computer Committee.
The following terms of reference were agreed:
Subject to the general authority of the Governing Board of the Institute, to
control the provision, operation and use of the NIRNS Atlas Computer
3 Developments to Date
The documents enclosed with paper NI/ACC/61/1 as well as NI/ACC/61/2
were taken as read. The Chairman referred briefly to the latest
position as reported in NI/ACC/61/3. Two points arose in discussion
of the basis for charging:
- The expected agreement with the Treasury that use by
university scientists should be free might be held to
leave out, for example, some university economics
research or data processing by educational departments.
It was agreed that the Secretary should either have the
words altered to refer to research use by universities or see that it was agreed not to interpret the words in
such a way as to exclude any university research suitable
for the Atlas.
- The proposal that use of the AEA and Government Departments
should be paid for was questioned, but it was agreed that
this was a finance matter which the Treasury had the right
to decide.
4 Proposal on the Operation of the Computer - NI/61/11
The following points were made in the course of a detailed
discussion of the proposals drafted by the Chairman and Dr Pickavance
in paper NI/61/11:
- In the light of later information reported in paper
NI/ACC/61/3, the estimated cost given at the end of
paragraph 2 of the paper should be amended to £3.4M,
and the estimated completion date should be given as
toward the end of 1963.
- The suggestion of a staff of 4 in the mathematics group seemed
low, but the division between the mathematics group and the
computing services group would depend very much on the head of the
Laboratory. The number 4 was suggested on the basis that
the mathematics group would be concerned with general development
and experimental work on the fullest use of the machine, and
not at all with programming particular problems. Even so, it
might be low.
- Some users would welcome the inclusion of a large group to work
on mathematical development suited to particular major uses.
The Meteorological Office for instance would like some basic
work to be done on the calculus of finite differences with the
object of improving the methods available for computer solution
of meteorological problems, thus allowing the meteorological
staff to concentrate on the physics rather than on the mathematics.
The general feeling of the meeting was, however, that a very
large staff would be needed to give an effective service of the
kind described, since many of the mathematical problems tended to
be different for each user. The Laboratory's mathematics group
would, of course, be interested in such problems and apply their
effort to the greatest general advantage, but it was thought that
it would be better generally for users themselves to undertake
the special mathematical developments for their particular
requirements.
- Similarly, in the case of data processing, it was thought most
efficient for users to deal with their own particular requirements.
Bubble Chamber users were quoted as a typical example of a group
who were already arranging to do so. It was important to arrange
that the computer took the problem as far as possible, i.e. to produce
a simple answer rather than to create a mass of results which required
further analysis.
- The total number suggested for the Laboratory was thought to be
about right. Professor Kilburn said that he expected to have a
total of 36 on the first Atlas at Manchester when it was in
operation (on a 24-hour basis) in about a year's time.
(About 5 to 8 of these would correspond as far as he could judge with
the mathematics group as proposed in our case.)
- The 5 scientific staff shown in the machine group included a
computer manager (a very responsible post), and 3 shift supervisors,
who would probably be in the Experimental Officer Class.
- With reference to the organisation proposed in the paper,
Dr Pickavance pointed out that it was proposed that the head of
the Laboratory would report directly to the NIRNS Atlas Computer
Committee, and not, for example, through the Director of the Rutherford Laboratory.
The proposal that administrative services
should be provided by the Rutherford Laboratory was questioned but
it was decided to leave points of this kind for consideration
after appointment of the head of the Laboratory. It was agreed
that detailed proposals on the delegation of financial powers
would have to be made. These would include certain powers for the
head of the Laboratory with higher limits after consulting
the Chairman of the NIRNS Atlas Computer Committee. Still larger
amounts would go through the NIRNS General Purpose Committee.
- Building plans were being prepared by the architects. Some
discussion with Ferranti's on the building requirements for the
machine had already taken place, and there was to be a further
discussion on 19th July. It was agreed that sketch plans would
be circulated to the Committee for comment. Accommodation for
about 30 visitors was planned and the Committee thought this
about right.
- After a discussion of the proposed Users Committee, it was agreed
that some such committee would almost certainly be set up by the
head of the Laboratory to help him to meet the users' needs to
the greatest possible extent. The sort of committee in mind
would be similar to the Rutherford Laboratory Visiting
Committee, dealing not with actual scheduling, but also
the arrangements for allocating time and for providing for
visitors. One difference from the Visiting Committee would
be that owing to the probably very large numbers of users,
the Committee would not include a large proportion of them.
Presumably, there would be arrangements for users with a special
interest at any particular time to attend a meeting by
invitation of the Chairman.
It was finally agreed that at the moment there was no need for
such a committee in addition to the Atlas Computer Committee itself.
It was agreed that for the benefit of the probably
large number of small users the procedure for requesting and
allocation of computer time should be simple and fairly cut-and-dried.
The proper route for any complaints concerning unsatisfactory
services would be first to the head of the Laboratory and, if the user was
not satisfied after this, to the Atlas Computer Committee.
- The question of the siting of the computer was considered to
be outside the terms of reference of the Committee, the Board
having agreed to the Harwell site after considering the many complex
questions involved.
- After discussion, it was agreed that it would be right to make
no charge for university research use of the computer. This
decision might have slight repercussions on university use of
the Manchester Atlas, for which Professor Kilburn said there
would have to be a small charge of the order of £50 per hour
to cover, for example, the cost of magnetic tape. It was agreed
that where the expense could be well forecast, there
should not be much difficulty in financing it. Mr Jolliffe
said that the DSIR had already made several grants
including provision for such expenditure.
- The full charge for time on the Atlas, including amortization of the
capital cost, was expected to be somewhere between £350
and £750 per hour. It was important to establish the charge in
good time, so that paying users, such as the AEA, could make
provision for the considerable sums involved.
In reply to a question, the Secretary said that he did not think
that the accounting arrangements would be such as to expose the
Laboratory to the temptation to allocate more time to paying
users when short of funds. He expected that any extra receipts
would go to the Treasury.
5 The Head of the Laboratory
The Chairman said that he thought that the key post of head of
the Atlas Laboratory should be advertised, and this was agreed.
After some discussion, it was also agreed that the post should be
one on the permanent staff of the NIRNS at the NIRNS equivalent of
the Civil Service Deputy Chief Scientific Officer.
The NIRNS follow AEA salaries and conditions generally. Posts
equivalent to the Civil Service DCSO are described as first
Senior staff appointments. The salaries are individually assessed
within a range starting at £3350. This is higher than the bottom
of the DCSO range because AEA and NIRNS staff have to pay superannuation
contributions. The Committee instructed the Secretary to
arrange for the advertisement of the post in this salary band.
The Chairman, Professor Kilburn, Professor Peierls and Dr Pickavance were invited to form a panel
to consider the
applications received and to make recommendations.
6 Forecast of Requirements for Atlas Time
It was agreed to defer consideration of quantitative forecasts of
requirements to the next meeting and it was recognised that such
forecasts would be extremely rough at this stage. Members who felt
able to make any forecasts before the next meeting were invited to
send them to the Secretary for circulation.
It was agreed that it would be premature to announce the
intended setting up of the Laboratory publicly, for example, by a
notice in Nature, in advance of approval by the Treasury, but
Dr Howlett and the Secretary were asked to keep the Inter-Universities
Committee on Computers, and the Royal Society Committee informed of the
position.
7 Future Meetings
The Secretary was asked to try to fix the second meeting
during the second half of October. the Committee would then consider
the frequency of meetings and try to fix dates for a year ahead.