June 04, 1996

Article at The Age

FEATURE | Monster From the Vault

Computer engineer Jurij Semkiw at the CSIRAC console in early 1960s [Museums Victoria]
Computer engineer Jurij Semkiw at the CSIRAC console in early 1960s [Museums Victoria]

WILSON da SILVA looks at a piece of world computer history about to take its place of honour

TUCKED AWAY in the dusty corner of a Melbourne warehouse is a little-known piece of world computing history. 

It weighs seven tonnes and took up most of a room when it was operating. That was back in 1949 – before the polio vaccine, before television in Australia, before you could even make a trans-Atlantic phone call.

Now, nearly 32 years after the vacuum-tubed CSIRAC computer was put in mothballs and largely forgotten, it is coming out again to reclaim its place in history.

For, as it turns out, CSIRAC – an archaic marvel of 1940s technology – has a hitherto unrecognised place in history: it is only the fourth stored-memory electronic computer ever built. It was built independently in Australia. And it is the only one of its generation still around.

“It is a marvel of engineering,” says Professor Peter Thorne, head of computing science at the University of Melbourne. As an 18-year-old student in the 1950s, Thorne started his career in electronics tending to the machine on weekends. “We didn’t realise at the time how advanced it really was.”

Its name conjures up images of a prop from a 1950s science fiction classic like Forbidden Planet. And it looks the part too. 

There is row after row of grey metal cabinets with dials and switches and gauges. Coloured lights are dotted in rows along its panels, and its innards is a jumble of thick wiring, mercury switches and vacuum tubes – 2,000 of them. 

Data was imputed on a ponderous grey metal console filled with toggles and switches and meters, with two large paper-filled spokes on either side. Here an operator would flick a series of knobs and run an application; in this case, in the form of a long spool of perforated paper. 

When fired up, the machine consumed some 30 kilowatts of power, and mechanical fans were needed to cool the cabinets lest they catch alight.

In modern terms, it had a mere 1024 bytes of random access memory, or 1k of RAM. Another 2.5 kilobytes could be stored on what may well be one of the world’s first hard-disks – a metre-long magnetic drum developed at the university that was used to temporarily store imputed programs.

Looking more like a power generator, the belt-driven disk drive had to be hand-started and would growl loudly when operating. Coupled with the roar of the cooling fans, and a mercury relay controller that had to be kept at 40 degrees Celsius, CSIRAC was not the most pleasant of working environments.

“In those days, everything was custom-made,” said Jurig Semkiw, one of the original CSIRAC engineers who, at the age of 24, helped develop the ‘hard disk’, and tended to the computer’s at times unreliable mercury relay lines, which had to operate at peak efficiency for the researchers to squeeze the 1k of RAM out of the machine. 

Semkiw, now 64, also learned to input programs in what might today be called “machine language”, the basic core instructions that drive a computer. “You had to be motivated enough to learn how to use it,” says Semkiw.

This slice of Australian history had, until recently, been forgotten. It was designed in 1947 by British-born radiophysicist Trevor Pearcey and developed with Australian electronic engineer Maston Beard at Division of Radiophysics, of what was then the Council for Scientific and Industrial Research, the CSIRO’s predecessor.

The “AC”, or “automatic computer”, followed on the footsteps of similar experiments going on in the United States. 

It was at the University of Pennsylvania where the world’s first truly modern computer, the ENIAC (Electronic Numerical Integrator And Computer) was born in 1946, based on earlier work dating back to 1939 at Harvard and Chicago.

Physicists and electronic engineers, having developed radar technology and complex algorithms for code-making and breaking, turned their minds to automated number crunching. Groups around the world attacked the problem independently, often with little knowledge of each other’s work.

This was at a time when Australia was isolated from the rest of the world, when visiting scientific meetings in the major capitals involved extended voyage by sea and when scientific publications arrived in Australia at least six months late.

Pearcey, a mathematician who had used large-scale computation for radar, teamed up with Beard, an engineer  experienced in vacuum tube-based pulse techniques, and built CSIRAC as a logical progression to experimental work in electronic logic components.

It embodied features that were considered novel at the time and operated more 1,000 times faster than the best mechanical calculators. And yet, it was less powerful than today’s cheapest pocket calculator and had less memory than a digital watch.

With this technological if now archaic wonder, the researchers may have scored a first – it is not yet known – by attaching speakers and programming the computer to play music. No-one now remembers the first song, but “The Good Ship Venus” and “Girl with the Flaxen Hair” were on the playlist.

As demand for computational power grew and with the arrival of transistors in the mid-1950s (which replaced the unreliable vacuum tubes) the Radiophysics engineers began to build in 1954 the more powerful SILLIAC at the University of Sydney. 

CSIRAC was donated to the University of Melbourne as part of a plan to have it utilised as a community asset. 

It tackled everyday problems like insurance risk analysis and calculating government drought relief programs. It also ran what may well be the world’s first computer-generated weather forecasts.

The restored colossus CSIRAC with some of the engineers who worked on it
The restored colossus CSIRAC with some of the engineers who worked on it

Next Thursday, computer scientists from around the country, in a gathering arranged by the Australian Computer Society, will come to the university to celebrate the 40th anniversary of its re-commissioning on June 14, 1956. Some will be from the original engineering team, including its inventors.

Organisers hope the meeting will draw out enough memories to compile a definitive history of early computing in Australia.

Some of those on the team, like Semkiw, have not seen the computer since he packed it away upon its decommissioning in November 1964.

“In those days, some people didn’t think there was a future in digital computing,” chuckles Semkiw. 

“We didn’t any understanding of how far it might go,” recalls Thorne. “But you really knew you were in the beginning of a revolution.”