Core blimey, what's going on!
When a computer went wrong some people
blamed hardware and some software.
In fact the reason used to be about
50-50.
Programmers usually blamed the hardware
and engineers usually blamed the software.
Hardware reliability and Windows 98
changed all that and now most, quite rightly, blame Windows 98.
Many years ago I was helping to commission
a new computer.
The machine was to form part of a demonstration
to the Australian Post Office.
Three of the computers were to be connected
in a ring and handle simulated phone traffic without falling
over.
The system was supposed to keep on working
if any single computer or even two computers failed.
The timescale for the demonstration
meant we had to work 24 hours a day (in two shifts) to finish
the hardware and software development.
I was one of two engineers on the engineering
shift and we worked our 12 hours testing hardware.
If anything needed sorting out on the
software front we had a programmer to help us. I recall that
he used a software language called "Minicoral". It
seems funny now, but it didn't seem odd at the time; if he needed
to do any major recompiling he used to go 200 odd miles to West
Drayton and do it using the UK "4-minute warning",
Air Defence System!
I suppose, because we'd supplied it,
we thought we were co-owners!
The other shift had two programmers
debugging software with an engineer to help them if required.
At the beginning of a 12 hour shift
you had to take over from the other team and at the end, hand
back over again.
At the end of a period of such intense
thinking and problem solving, one's brain got a bit worn out.
The effect was peculiar. Near the end of a shift, say at five
or six in the morning, one could see things really clearly and
solutions to problems used to be really obvious. Unfortunately,
in reality they weren't and it was just because ones brain was
worn out.
We had a log book and at the start of
a session, when you read what you'd written at the end of the
last shift, you couldn't imagine how daft you'd been.
One morning when myself and my fellow
engineer arrived we found the two programmers rummaging around
inside the computers memory, watched with interest, by their
engineer.
The computer had a "core store"
which used miniature ferrite rings on which to store a bit of
data.
The store was 24 bits wide and 4k deep.
In modern parlance that's 12kbytes.
The core store occupied a 19" rack
mounted unit about 12" high which weighed about 60 pounds
and most likely would have been worth a years salary. There were
a total of 96,000 ferrite rings in the unit and these were held
in an incredibly densely packed module which looked a bit like
a 3-dimensional spiders web in the shape of a 4" cube.
One of the programmers had the ferrite
ring module on the bench and he had a soldering iron in his hand.
We were flabbergasted.
"What on earth are you doing",
I asked.
"Oh", he said, not looking
up, "We reckon two of the bits are the wrong way round and
I'm just rewiring them".
Of course he'd had a program fault and
he'd blamed the hardware.
We had to get a new core store before
we could start our shift as he'd wrecked ours.