1959-1961: Upgrading AEA and NIRNS Computer Power
See NRDC and the case for a British supercomputer, 1956 - 1960 for more detail.
The need to update the computing power available to the nuclear research community in the UK
proceeded in the context of two announcements:
-
Univac LARC: designed to meet the requirements of the Livermore Research Laboratory defined in 1956, anticipated delivery 1958, installed in 1960
-
IBM STRETCH (7030): designed to satisfy the requirements of Los Alamos defined in 1956, first STRETCH installed there in 1961
Both computer systems were supercomputers,
promising to be 100 times faster than existing computers. This opened up the possibility of
developed programs to tackle problems clearly outside the capability of existing computers.
STRETCH was of particular interest due to close collaboration between Aldermaston and Los Alamos
and IBM's intention to market it as a product.
- January 1957: AERE send Report to NRDC
- By 1960 the AEA will need a computer comparable to STRETCH.
- March 1957: NRDC Proposal
- Halsbury recommends to NRDC Board that they spend £1M over 4 years
to develop a supercomputer. Agreed if AEA agree to buy one, NRDC to design the machine
and place contracts for its implementation.
- April 1957: First Harwell Computer Conference
- AEA needs a very fast computer with large storage capacity to solve 3-dimensional field problems on top of the needs of the weapons programme.
- May 1957: AEA propose to purchase an IBM STRETCH for Aldermaston
- June 1957 - January 1958
- Several attempts by NRDC to put structure to its consortium proposal but with little enthusiasm from AEA.
- February 1958: Second Harwell Computer Conference
- Harwell put forward their requirements for a supercomputer
- Manchester University outline their plans for their MUSE Project
- March 1958: Ferranti announce the development of Orion
- Agreed to cooperate with Manchester University on their development of MUSE.
- March 1958 - October 1958: NRDC
- Various attempts to organise the development of a supercomputer under NRDC's control with no real support from AEA, Ferranti or Manchester University
on the proposals.
- October 1958: Ferranti to Develop Atlas
- Atlas would be based on MUSE and the project run jointly by Ferranti and Manchester University
- Sebastian de Ferranti jokingly thought it should be called BISON: Built In Spite Of NRDC ;-)
- March 1959: Ferranti Presentation of Orion and Atlas
- Orion would be comparable in size to an IBM 7090 (3 to 5 times the speed of Mercury).
First customer delivery expected in June 1961.
- Atlas would be essentially the MUSE machine designed by Manchester University.
First customer delivery expected in 1963.
- October 1959: AEA Computer Policy Committee
- Considerable growth in computer power needed for reactor design and CTR experimental design.
- Hire or purchase an IBM 7090 for Risley now, and order an Atlas for delivery to Harwell in 1963.
- An Orion be purchased for the use of RHEL, Culham and Harwell for delivery in 1961.
- Aim that the Atlas programming system is compatible with that of the 7090
- January-June 1960
- Harwell had meetings with Ferranti and interested parties in putting together the case for the purchase of an Atlas;
also several meetings of AEA CPC.
- November 1960: AEA Meeting
- Approved 3 recommendations of CPC:
- IBM 7090 should be purchased for Risley when Stretch installed at Aldermaston
- Defered the decision on a KDF 9 for AEA Winfrith
- AEA supported the collaborative purchase of a Ferranti Atlas to be used by AEA at Harwell and Culham
- December 1960: Proposal to Treasury
-
- A.E.A. should order an Atlas from Ferranti
- The machine should be installed at Harwell
- Half its time would be sufficient for Harwell and the overflow needs of Culham, Risley and Winfrith
- A substantial amount of time should be allocated to RHEL and universities
- 1960: Authorisation
-
- An Atlas should be ordered by AEA
- NIRNS should run the machine
- Equal shares to NIRNS (RHEL), Harwell and Universities/Government Departments
- NIRNS and university users should not be charged. Other users should pay.
- January 1961: NIRNS
- NIRNS were made aware of the proposals in January 1961:
- Aldermaston to rent an IBM 7030 for 3 years (1962-1965). The lowering of the anticipated 7030 speed
plus security considerations indicated that time on the 7030 would not be available to academic institutions.
- If NIRNS agreed, Harwell would purchase a large Atlas and install it at Harwell under NIRNS control
with a minimum allocation of time to Harwell to be agreed
- The NIRNS charter indicated it was primarily organised to provide and operate large nuclear science facilities
for academic users. It was believed there was no legal objection under the terms of the Royal Charter for NIRNS to manage
a computer facility with a significant nuclear science load.
- February 1961: NIRNS
- NIRNS agree that an ATLAS should be installed at Harwell under their management and control:
- The order should be placed quickly
- Computers of KDF9 power should be installed at some universities as well
- One or more universities may require Atlas-power computers of their own by 1964
- March 1961: NIRNS
- The decision was made that the new Atlas Computer Laboratory should be a separate
laboratory between Harwell and RHEL and that some arrangements should be made with
RHEL regarding support services that might best be provided by RHEL for both Laboratories
- 1961: AEA
- A letter of intent was placed with Ferranti to purchase an Atlas computer to be installed at Chilton
close to Harwell. Much of 1961 was spent in specifying the details of the formal order
with acceptance tests.
- August 1961: London University order an Atlas in partnership with the British Petroleum Company
- 1961-1964 NIRNS
- NIRNS oversaw the creation of the Atlas Computer Laboratory. Details
can be found in the Minutes
of the NIRNS Atlas Computer Committee
- 1965-1971 SRC
- The SRC Atlas Computer Committee took over from the NIRNS Committee when the Science Research Council
was formed in 1964.
Details
can be found in the Minutes
of the SRC Atlas Computer Committee that ran from 1965 to 1971.
- 1972-1975 SRC
- The Reconstituted SRC Atlas Computer Committee took over from the SRC Atlas Committee
when the decision was made to restrict the Atlas service to SRC Grant Holders
giving them a guarantee of time over the period of their grant.
Details
can be found in the Minutes
of the SRC Reconstituted Atlas Computer Committee that ran from 1972 to 1975.