Information Technology Reference
In-Depth Information
open-ended architecture with minimal hardware based around ferrite core memory
and related read-only store. They dubbed the “blue sky”, unfunded project
CIRRUS
,
and proceeded to produce a detailed design. The instruction microprograms were
simulated by Trevor on
CSIRAC
and software design started [47].
Adelaide University saw a very promising future for the design, supported the pro-
ject and stimulated funding from the Postmaster General, and Weapons Research
Establishment.
The basic hardware elements that Murray used in
ADA
and
SNOCOM
were extended
for
CIRRUS
and constructed by a PMG contractor. As the hardware developed it be-
came clear that it would be so fast that a form of multiprogramming would be needed to
keep the processor busy. The team knew of a few commercial examples (
Honeywell 800
and
IBM 7030 Stretch
) but their requirements were different. John Penny wrote a simu-
lator on WRE's
IBM 7090
and developed the operating software there [48].
CIRRUS
came into operation in late 1963 with 4 user workstations. It was fast, in-
expensive, and well ahead of commercial contemporaries. It was heavily used despite
the installation of a CSIRO
CDC 3200
and a
CDC 6400
. As well as teaching and
research programming, special purpose on-line control and signal-processing worksta-
tions were built. A prolonged breakdown in 1969 triggered the purchase of a
Data
General Nova
, and
CIRRUS
was taken out of service late in 1971 [49].
CIRRUS
is preserved in Adelaide Uni's Electrical Engineering Department.
Australian industry did not take up this cheap and flexible design and Adelaide Uni
did not pursue computer development. Murray Allen spent a year with Control Data
in the USA contributing to the
CDC 3000
series computers, then he moved to the
University of NSW. Trevor Pearcey and John Penny joined the new CSIRO Division
of Computing Research.
11 ATROPOS
The Weapons Research folk needed to know where the rockets they were testing were
likely to come down. They had an “impact predictor” system to do this
12
, but with the
1958
Blue Streak
project the impact point would shift by 60 km every second, and
they decided a digital system was needed. It had to run the 5,000 instruction proce-
dure five times a second and there wasn't a commercial system WRE could afford
which would do that [50]. A small team of WRE engineers (Ian Hinckfuss, Ron Keith
and Ian Macaulay) visited the UK in 1957 and included RRE where Trevor Pearcey
was working on
TREAC
[51]. This machine was fast, with parallel operation and very
good arithmetic [52]. They felt they could do even better with transistors instead of
valves, and core memory instead of William's Tube CRTs.
They had a design by 1960 and WRE workshops constructed their computer by the
end of 1962. This was large for a second generation machine: 6m wide, 2m high and
½m deep. Transporting it 150 km from the labs near Adelaide to the Woomera Range
was done on an air cushioned truck, at 15 km/h.
Its major input was from two radar units 200 km away, so they also had to invent a
reliable data transmission system [53]. Its job was to convert radar data to position
12
Rocket tracking radar was linked to plotting tables watched by the Safety Officer.
Search WWH ::
Custom Search