By 1961, the plan for an Atlas to be installed somewhere on the Chilton/Harwell site had been agreed. The case was made on the basis that the customers would be:
By mid-1961, the decision had been made to set up the Atlas Computer Laboratory as a new Laboratory on the Chilton/Harwell site run by NIRNS. It had been ascertained that NIRNS could legally run a computer service free of charge to universities and was not restricted to providing particle physics support.
By now the Aldermaston machine had become an IBM 7090 and the relative computer power had changed as shown below. This attempts to take into account the allocations of time provided by Aldermaston to other organisations.
Almost as soon as the decision had been made to purchase an Atlas and run it under the control of NIRNS, additional bids were being formulated. Both London and Cambridge saw use of a NIRNS Atlas as a stepping stone to acquiring their own systems. Harwell were unsure that their requirements would be met by a NIRNS Atlas. Rutherford required on-line connections to their measuring devices for particle physics tracks and were concerned as to whether this would be feasible with a machine controlled by a separate organisation. Culham was interested in acquiring their own computer system and Winfrith wanted an upgrade to their ageing Mercury.
The case for the Atlas was based on a variety of demands:
| Organisation | Atlas Fraction |
|---|---|
| Harwell | 0.15 |
| Culham | 0.10 |
| Rutherford | 0.10 |
| Government Departments | 0.10 |
| London University | 0.10 |
| Cambridge University | 0.10 |
| Other Universities | 0.10 |
In consequence, any one customer going away would not significantly alter the case especially as the machine would not be delivered for another 2 years when demand would almost certainly have increased.
By July 1961, it was clear that London University would purchase its own Atlas starting with a basic configuration and then extending it. It would be purchased partly by funds from BP in return for computing time, partly by grants from the University and the UGC and partly by borrowing money. By selling computer time, it aimed to pay back the money borrowed. Thus the London Atlas would never provide anywhere near an Atlas unit of computer power to the academic community and would not seriously alter the case for the NIRNS Atlas.
