At the January 1967 Meeting of SRC's Atlas Computer Committee the discussion started with regard to the future hardware and role of the Atlas Computer Laboratory:
The demand from the government and academic sectors clearly indicated the need for more top-end computer power.
The Ministry of Technology believed that the new proposed ICT 1908 (sometimes called P.51) with up to 4 processors would meet these requirements and urged for a buy British policy. This was an ambitious project with the operating system spread across the four processors
The University Science and Technology Board (UST) revised their Forward Look to include it as neither SRC had the funds to purchase an Atlas replacement. The Atlas Computer Laboratory was a Centre of Excellence and its facilities must be kept up-to-date.
The main alternative from the USA was the unannounced CDC 7600 In late July, Bob Hopgood visited CDC in Minneapolis and Chippewa Falls to get an In Confidence briefing on the CDC 7600 that was reported back to Jack Howlett, Lord Halsbury, Christopher Joliffe and Brian Flowers.
By November, it was clear that the only machines of sufficient power to make a meaningful replacement for Atlas were the CDC 7600 and the 4-processor ICT 1908.
UST believed that ordering a ICT 1908 as a replacement for Atlas would have an important effect on the attitude of other potential buyers to the machine, at home and abroad.
Discussions with CDC suggested that they would be happy to see a UK machine front-ending the CDC 7600.
ICT by now was close to a merger with EEL to create ICL and still had not provided a specification of the ICT 1908 (P.51). The original plan for a 3-processor system upgradeable to a 4-processor system had been changed to a 2-processor system.
The Atlas Computer Committee suggested that an ICL 1906A (currently the top-of-the-range for the 1900 series) could be ordered with a second one rented until the P.51 arrived.
By July 1968, the P.51 had been renamed the ICL 1908A. ICL could deliver either a 1908A+1908A system or a 1906A+1908A system. The first 1906A would be available in June 1970 and first 1908A in April 1972
UST recommended that an order for the ICL 1906A+1908A system be placed and this was agreed.
Preparation for a new building to house the ICL 1906A+1908A was started. On 11 July 1969, ICL abandoned the production of the 1908A. As the 1906A+1908A system was to be placed in several university sites as well this was very bad news.
Rather than attempting to keep both the System 4 and 1900 Systems in production, ICL's aim was to produce the first system of the New Range at the top of the range and use the Manchester MU5 Project as the architecture model for the New Range.
Abandoning the ICL 1908A left the whole university procurement strategy in a mess. The aim had been to install 6 to 8 1906As in universities with the Atlas Laboratory's 1908A being responsible for running those large jobs that would disrupt the 1906A services.
By December, it was decided to continue with the purchase of the 1906A with the size of the main store increased from 128 K to 256 K. It would help the effort available to get the university 1906A systems into operation. Again, it was suggested that the ICL 1906A should act as a front-end to a future large batch processor probably of USA origin.
A separate Engineering Board was set up by SRC recognising the growth in engineering research.
Much of 1970 was spent in trying to redefine the role of the Atlas Computer Laboratory. Discussions continued with CDC concerning a possible installation of their new supercomputer, the STAR-100, that would replace the CDC 7600. The Ministry of Technology stated that they would look very hard at a proposal to put a big American machine at the Atlas Laboratory. They would want to know what work would be done on it and why an American machine was necessary (they conceded that they had let London University purchase CDC equipment). On the other hand, the Computer Board thought the only viable national role for the Laboratory was at the top of the university computing hierarchy.
The Nuclear Physics Board of SRC agreed to upgrade RHEL's IBM360/75 to the IBM 360/195 to be delivered in 1971 and 20% of the machine would be available to the Atlas Computer Laboratory to support SRC users.
By November, it was agreed that the role of the Atlas Laboratory should change and in future it would just support SRC-funded projects.
An IBM 1130 was purchased by Atlas to provide their users access to RHEL's new IBM 360/195. An Opening Ceremony for the ICL 1906A was planned for 1972. The building plans were revised as the 1906A would take up less space than the combined 1906+1908A. The SRC approved projects would be guaranteed access to appropriate computing time. Discussions continued with CDC and ICL with regard to a replacement for the 1908A. The CDC STAR with its special vector-processing operations was an interesting possibility. The Ministry of Technology continued to state that purchase of a big American machine would be in direct opposition to the Government's stated policy of acquiring large computers from ICL unless strong arguments existed to debar this.
The main possibilities for increasing the Laboratory's computer power were the new ICL P4, a CDC 7600 or a CDC STAR-100. The CDC 7600 offered the most straightforward and least challenging solution. The CDC STAR and the ICL P4 both had unproven novel architectures. In April 1972, CDC offer a STAR-100 for delivery in 1974 with a performance 100 times greater than Atlas at a cost of £4M.
A Reconstituted Atlas Computer Committee (RACC) was set up by SRC to advise the Science Board on the operation of the Atlas Computer Laboratory in its new role and, in agreement with the other Boards, on priorities for the work of the Laboratory.
There had been little growth in SRC's funds over the previous few years. UK's subscription to the CERN budget fluctuated yearly and tend to adversely effect the funds available elsewhere. The Advisory Board for the Research Councils (ABRC) made the decision to keep SRC's funding down relative to the other Research Councils. ABRC wanted to reduce the expenditure on the sciences that are extremely expensive per researcher in order to sustain growth elsewhere. In consequence, SRC could not fund all the large science activities that the SRC Boards wished to pursue. The forecast for 1974/75 would be reduced by £3M. The overall budget for 1973 to 1977 instead of an increase from £145M to £162M became a decrease to £133M
SRC's Council Meeting in March 1973 discussed the regrouping of activities in the Council's establishments. In anticipation of future changes, the Rutherford High Energy Laboratory was renamed the Rutherford Laboratory, the Daresbury Nuclear Physics Laboratory was renamed the Daresbury Laboratory and the Radio and Space Research Laboratory was renamed the Appleton Laboratory.
RACC discussed a proposal from ICL to install an ICL P4 at Chilton. The Committee supported the proposal from their Panel:
Much discussion around a number of proposals:
The final decision was:
For a variety of reasons, the 7-year discussion of a major replacement for Atlas had come to nothing but an ICL 1906A, twice the power of Atlas and a 20% share of the Rutherford IBM 360/195.
