The policy for the purchase and management of the ATLAS computer was formulated during the first half of 1961, and received Treasury support in July. The entire project has been transferred to the National Institute for Research in Nuclear Science, who will set up at Harwell a new Laboratory called the ATLAS Computer Laboratory. this will provide a service to the Institute, the Atomic Energy Authority (primarily Harwell and Culham), the universities and Government research establishments. the buildings will be outside the AERE security fence, and only unclassified work will be done in the Laboratory. in conformity with the general policy of the National Institute, no charges will be made to the universities; other users will be charged for the time they use, and the Laboratory is authorised to sell any surplus time to industry and commercial users at rates which have yet to be settled.
The Laboratory is managed by the following Committee which, for this particular project, exercises the authority of the National Institute's Governing Board:
(* A member of the Governing Body of the Institute)
The present plan for staffing the Laboratory is that it should consist of a Director, an Operations Group of from 23 to 33 according to the number of shifts worked, a Programme Group of 10, and a Mathematics Group of 6. The building will provide rooms for members of universities and similar institutions who would like to work for a time in the Laboratory whilst developing a programme for the machine, or carrying out some large computational project. The Committee advertised the post of Director nationally, and from the applicants selected J Howlett, who formally took up the appointment on 1st December 1961.
The specification for the installation is as follows:
A formal contract is still being negotiated with Ferranti Limited, but they have a letter of intent to purchase the machine at a fixed price of £2.55M. the financial arrangements include an undertaking to pay part of Ferranti's development costs; the maximum payment here would be £700,000, to be reduced at an agreed rate for every other ATLAS sold. Another machine has, in fact, already been ordered by London University, and consequently the payment will not exceed about £550,000.
Manufacture of the Harwell machine started in December 1961, and is scheduled for completion by October 1963, when the whole machine should be assembled in the factory. Ferranti say that at this stage they cannot estimate at all accurately how long it will take to work the machine up to full reliability, and they have programmed nine months for this, giving July 1964 as the date for starting installation at Harwell; the expectation is that the machine will be working on the site by the end of 1964. Progress will depend very much on what happens at Manchester University; their machine is due for completion in August 1962 - except for the punch card input/output which will be added before the end of the year, and at the moment is not more than two weeks behind schedule. The core store has been commissioned in the Plessey factory; 8K will be installed at Manchester by mid April and the second 8K by mid May. Professor Kilburn has reported that the work already done on the machine shows that the arithmetical speeds will certainly be up to expectation.
