ACD Atlas Computing Division PERQ History: Part VII

Jump To Main Content

Jump Over Banner

Home

Jump Over Left Menu

1983

34. HARDWARE

34.1 Introduction

The year started with the original PERQs as the only type available but with a strong belief that the KRISTMAS developments (Section 20.2) would produce a new PERQ (later called PERQ2) during the year. SERC had an outstanding need for upgrades to the I/O Board to drive the Cambridge Ring interface and 16K WCS in place of 4K for a number of users who wished to re-microcode the PERQ for particular applications particularly in the AI area.

34.2 PERQ2

The PERQ2 was a completely re-engineered version of the original PERQ (later called PERQ1s). The disc had been replaced by a smaller, less noisy 8 inch 35 Mbyte Winchester. Memory sizes were 1 or 2 Mbytes. The system was completely repackaged in ICL-designed covers and colours. The PERQ2 could have either an A4 or landscape display. The landscape display would have the same pixels/inch resolution as the A4 display making it significantly superior to any of its competitors (and it stays that way today). The keyboard had been replaced by a much more robust industry standard one and the bulky Summagraphics Tablet had been replaced by a small A4 tablet and mouse. Unlike the PERQ1, the only writable control store option was 16K.

The arrival of PERQ2 meant that ICL would target any new developments at it and would not feel constrained by the limitations imposed by PERQ1. Thus, for example, PNX and GKS were targeted at a 1 Mbyte 16K WCS system.

In March 1983, Three Rivers launched the PERQ2 in the USA. ICL were unhappy about one or two safety requirements and delayed their launch until the Autumn when it was officially launched at SICOB in September.

ICL were worried about tolerances on some of the board tracking. As a result, ICL reworked both the 16K WCS and the I/O Board.

ICL loaned SERC a Three Rivers PERQ2 for a month in March in order that some benchmarking could be done. This confirmed small improvements in performance (10 to 20%) in the disc accessing but, in general, the machine was a re-engineering of the original design and it performed much the same as the PERQ1s.

It was originally intended to put a second 35 Mbyte drive in the PERQ2 cabinet so that a much larger system could be obtained. However, with the eight inch discs (later called PERQ2/T1) this proved undesirable and a PERQ2/T2 with five and a half inch Micropolis drives was produced which could have two disc drives.

ICL planned to manufacture about 4 systems in July, 50 in August and 100 in September. These dates slipped a bit but not significantly.

ICL were anxious to run down production of PERQ1s as soon as possible and concentrate on tooling up for PERQ2 production. RAL ordered a batch of 15 PERQ1s in February for grants awarded in December 1982 and spent most of 1982 attempting to get SERC to agree to a bulk purchase of PERQ2s so that early delivery slots of the new system could be obtained. Unfortunately, due to internal wrangling within SERC, to be described later, no early orders were placed for PERQ2s and consequently they were late in appearing within the SERC environment (first quarter 1984).

34.3 Colour

Three Rivers launched an add-on Metheus colour system in the USA but ICL spent a great deal of time assessing alternative suppliers. One possibility was to wait for the PERQ-DAP which would have a quality colour monitor as an option. Both the PERQ-DAP and an add-on colour display were likely to be products at some stage.

34.4 16K WCS

The 16K writable control store was available in 1982 and ICL had decided to buy these in from Three Rivers for the PERQ1, leaving ICL's main thrust at redoing the board for the PERQ2s.

ICL eventually decided to make PNX only available on 16K WCS machines. As a result, SERC had a major upgrade to do which, due to negotiations being protracted, was not completed until much later.

34.5 PERQ3

By February 1983, the design for PERQ3 was started. The decision was made between Three Rivers and ICL that this would be a range of processors with the low end system being 68020 based. The aim was to provide a system with the ability to have multi processor configurations with specia1ised processors such as the PERQ DAP and a bit-slice processor coexisting with the low-end processor in a flexible multiple processor system.

34.6 Summary

ICL's hardware developments in 1983 were mainly aimed at the launch of the mid-term boost system, the PERQ2. As a result, activities around PERQ1s were less fast than one would have hoped. ICL's strategy did mean that the original PERQ1s were becoming obsolescent by the end of the year.