?

Log in

No account? Create an account

Fri, Jul. 23rd, 2004, 09:54 pm
We love you, CSIRAC!

While I was visting Ristin, he took me to the Melbourne Museum. There's lots of cool stuff there and it was a good opportunity to learn about Melbourne's history, but for me the highlight of the trip was CSIRAC.

CSIRAC was Australia's first computer, and only the fourth programmable computer in the whole world. Development began in 1947, and it was in service at CSIRO until 1956, after which it worked at Melbourne University until it retired in 1964. Computing was a whole new field back then, so it was basically designed from the ground up. This machine was a crunching powerhouse of the day, performing at a blistering 1 000 operations per second (.001Mhz). It had a memory capacity of 768 (20 bit) words (about 1.8KB in modern terms) and an interface consisting of oscilloscopes, blinking lights, a speaker, and punched paper tape, which may have made it a bit limited for games (although some games were developed for it). It used mercury tubes for memory. Part of what I find so fascinating about CSIRAC is the challenge it must have been to program it in its own arcane language, trying to make the most out of every bit.

After its well deserved retirement, CSIRAC was delivered to the Melbourne Museum; the only computer of its generation to still survive intact. CSIRAC was one of Australia's most important technological achievements, back when we still believed we could do anything. Well may we say, "We love you, CSIRAC!"

Fri, Jul. 23rd, 2004 01:10 pm (UTC)
darth_phylos

I thought CSIRAC was the fifth programmable computer? I love the old school computers, and was looking into them a while ago for a sci-fi story I was going to write. I even learnt Turing's programming language.

Fri, Jul. 23rd, 2004 04:11 pm (UTC)
marko_the_rat

The Melbourne Museum called it the fourth. Maybe it depends on what you call programmable?

Fri, Jul. 23rd, 2004 04:36 pm (UTC)
darth_phylos

I can't remember where I read that it was fifth - it was a while ago I was looking into stuff like this.

Actually, I just followed your link, and it says fifth as well.

Fri, Jul. 23rd, 2004 05:33 pm (UTC)
porsupah

I don't suppose the museum listed its "predecessors", did it? I'm wondering whether Colossus was included, which was unrecognised for quite some time after its active service, being lost amidst secrecy.

Of course, even now, you can gain something of a feel, in a very different way, of the benefits of truly knowing a processor's architecture. sim_g4 was a nifty processor timing simulation tool from Motorola; with it, you could see, per tick, at what the stage of execution of the two instructions on the move was. That rendered detection of cache misses trivial, as you'd have an instruction stuck for eighty ticks or more while the cache did its thing. Execution unit contention was also made plain. All such problems can be circumvented, if you know the color of pixie dust to scatter upon the source. ^_^ It led, for me, to quite a good understanding of precisely how well instructions could move through the pipeline, and fit in nicely with my fondness for efficiency in coding. (And then came the 2001 market crash, prompting our new parent company to execute us in a different sense)

Fri, Jul. 23rd, 2004 05:48 pm (UTC)
marko_the_rat

I quote from The Last of the First: CSIRAC: Australia's first computer by Doug McCann and Peter Thorne:
CSIRAC was probably the fourth or fifth electronic stored-program computer in the word to run a program. It is difficult to definitively assign a ranking to the operational dates of many of these first generation computers because much depends on one's definition of 'operational'. By the term 'operational' do we mean when the first test program operated or when the computer commenced routine operation?