The initial order from Three Rivers had given us 11 systems of which 5 were owned by ICF and 6 by DCS.
The position in mid-September 1981 was that we had approval to purchase:
| Interactive Computing Facility | 10 |
| Distributed Computing Systems | 6 |
| Software Technology | 10 |
| Daresbury Laboratory | 1 |
| TOTAL | 27 |
The Computing Section at Daresbury had requested RAL to order a system. They were interested in doing the development necessary to interface it to CAMAC and were also interested in Local Area Network developments.
There were other bids in the system including ones local to Divisions in RAL and PERQs that had been awarded on grants but for which authorisation had not been received.
The UK price of the PERQ for a 24 Mbyte disc system was $52,000. The agreement with ICL gave SRC discounts as follows:
| 1 to 10 units | 10% | $46,800 |
| 11 to 25 units | 15% | $44,200 |
| 26 to 50 units | 20% | $41,600 |
| Over 50 units | 25% | $39,000 |
There was still discussion as to whether another level of discounting could be obtained past the 50 mark.
There was still no large scale authorisation to purchase in bulk. Consequently orders to ICL had to be made two at a time during September.
The aim was to have a bulk purchase from ICL and then sign off systems to individual projects within SRC as they requested them. The hope was that at the launch, ICL would be able to announce an order for 20 systems in addition to the 11 already purchased and that SRC would order a further 20 in the near future. ICL, on their side, committed to manufacture in the UK Letchworth factory early in the New Year. This was all given in the Press Release dated 29 September 1981.
Eventually, the first 27 systems were ordered in pairs and it was not until February 1982 that a bulk order for 16 PERQs was made. By the end of the SRC financial year (31 March 1982), 56 PERQ systems had been ordered and delivered by ICL. The funding for these was as follows:
| Interactive Computing Facility | 10 |
| Distributed Computing Systems | 6 |
| Software Technology | 10 |
| Daresbury Laboratory | 1 |
| Central Computing (for assessment) | 3 |
| Hatfield Poly CCSC Grant | 1 |
| Edinburgh AI EMR(ICF) | 1 |
| Central Office EDPU | 1 |
| ICF | 3 |
| EB Pool | 13 |
| SB Pool | 3 |
| NP Pool | 4 |
| TOTAL | 56 |
In the initial phase, many of the systems were used by the team of people working on the project for systems development, support and demonstrations. ICL were attempting to sell every system that they could manufacture and, consequently, some systems were also used by the ICL staff doing development work. Between 20 and 30 systems were used in this way at RAL. The need to support the existing operating system, the development of new operating systems etc meant that a large number of systems were needed even when the work force was relatively small, to avoid dumping and reloading different versions of the operating system. There was also a considerable interest by external people in loading specific programs to see how they performed in this new environment.
The build up of systems near the end of the financial year was quite large. The installed base of PERQs each month was:
| September 1981 | 11 |
| October 1981 | 11 |
| November 1981 | 11 |
| December 1981 | 17 |
| January 1982 | 23 |
| February 1982 | 45 |
| March 1982 | 56 |
The large number of systems arriving in February and March put an unreasonable load on the person checking systems, loading software on to them and generally checking them out. Consequently, even by the end of May, 11 of the systems were still not unpacked and available for use.
The original PERQs were ordered as 0.25 Mbyte systems. By March, ICL had decided not to make such systems as it was cheaper to make 0.5 Mbyte systems. Consequently, the systems actually delivered in February and March were a mixture of 0.5 Mbyte and 1 Mbyte systems.
ICL were interested in making the minimum configuration a 0.5 Mbyte system as it was already becoming clear that the performance of the 0.25 Mbyte systems under the POS operating system was significantly lower than the 0.5 Mbyte system. As a result it was decided to upgrade those systems that only had 0.25 Mbyte until all systems had a minimum of 0.5 Mbyte.
By April 1982, the price of a 0.5 Mbyte system to SERC was £19,950 and a 1 Mbyte system £22,150.
By October 1982, 90 PERQs had been ordered and 88 delivered. Of these 90 systems, 62 had 1 Mbyte boards, 19 had 0.5 Mbyte boards and the remaining 9 had 0.25 Mbytes. A large number of systems had been upgraded and the aim was to upgrade the last 9 systems to 1 Mbyte boards as soon as possible.
The change of memory size was not a straightforward board swap and time was required to make each change. Also, 6 hours of work was needed to unpack a PERQ, test it, format the hard disc, write the software, and re-pack prior to delivery. Consequently, a great deal of activity took place in this area during the first nine months of the year.
As early as October 1981, Science Board had started receiving requests for PERQs. Prof Leadbetter at Exeter was attempting to change his LSI-11s to PERQs. Owen Mills at Manchester had put in a grant concerning the DEC10 crystallographic database.
By October 1982, the four Science Board funded PERQs were being used as follows:
Of the remaining systems purchased, one was owned by SB and that was placed with Prof J N Murrell at Sussex. A Central Computing PERQ was loaned to Owen Mills at Manchester University to allow software relevant to the Science Board to be moved off the DEC10. Consequently, 4 out of 5 SB systems were effectively being used on SB activities directly and the fifth as systems support. This compares with the Engineering Board where about 40% of the systems purchased were being used in software development.
PERQs were installed in 41 different application sites by October 1982.
