The 1906A operating system was called George 4. This was a version of ICL's standard George 3 operating system adapted to take advantage of the paging hardware. The system includes good filestore control with incremental backup. A multi-access system, MOP (Multiple On-line Processing) was being developed to allow terminal connections to the 1906A rather like the Sigma 2 facilities on Atlas. File editing, batch submission and the ability to interrogate the status of jobs were the main features.
On its arrival, the 1906A ran the George 3 operating system with no MOP support or the operators' control system called GERONIMO.
By September 1971, the hardware of the 1906A was working reasonably well under George 3 (Mark 5.4) and the aim was to start a restricted internal service before the end of September. The paging hardware had not been delivered and significant modifications were needed to the installed 1906A to fit it. Effectively, most of the CPU needed to be replaced by a paged version. The old system would then be refurbished for the next customer waiting for paging hardware. The changeover would take 8 night shifts to complete and this was not completed until January 1972.
In February of 1972, George 3 Versions 6.2, 6.3, 6.4 were installed. An attempt to install George 4 was made in March. Placement of standard software such as libraries was found to be a problem. If made part of the filestore on the permanent disc, they were being dumped to tape before their next use. The solution was to mount them on EDS30 removable discs. Thus the EDS30, removable discs, were used for permanent libraries to ensure they were not removed!
The George Scheduler was in two parts with the High level Scheduler being designed for local modification so that an organisation could define its own scheduling strategy. Much effort was put into instrumenting the performance so that informed changes could be made and their effect measured.
Communications initially was through ICL 7020 remote job entry stations directly attached to the 1906A but the aim was to go to a remote job entry station that could be attached to either the 1906A or the 360/195 via a front-end processor attached to both computers.
A 60 small jobs benchmark was set up that ran on Atlas in 8 minutes. This initially took 49 minutes on the 1906A but tuning soon brought it down to 16 minutes.
George 4 was eventually brought into operation in June 1972. The Geronimo system arrived in the summer.
Performance of the system was not good and most bugs were eventually being left to Version 7 of George 4 that was due for arrival near the end of 1972. In fact it was mid-1973 before George 4 Mark 7 was introduced and performance of the machine began to improve. This coincided with the introduction of the EDS-60 filestore. By now, the GERONIMO system was working well and the operators had much better control of the machine. By the end of 1973, George 4 was reasonably stable and the activities switched to improving the system as a whole.
