was a machine!
by FRED GRUENBERGER
The CPC—there was a machine. In fact, for a long while, there was the machine. It wasn’t what you’d call today a computer (what you’d call it today would be a mess), but in its heyday, around 1950, there were 700 of them and they were doing most of the computational work in the world.
It is difficult even to describe the beast to today’s students. There was this tabulator, see, and a card punch, and a wired electronic calculator, and some huge mechanical storage devices …all connected by huge rubber-covered cables all over the floor. Top speed in floating point was 150 operations per minute, and no loops or subroutines or macros. In order to get answers, pal, you had to be clever.
There was another feature, though, that we lack today. The CPC had sportsmanship. When your program didn’t work, there was always a sporting chance (p = .50) that it was the machine that was at fault. This made for a more lively atmosphere. None of this dull feeling one gets when the machine never goofs.
The brainpower that today is spent on devising better monitors and more efficient compilers was applied in those days to the construction of plugboards. Decision logic had to be implemented through relays, of which there were never quite enough. Here was a fertile field for ingenuity, and people dreamed at night of creating
scheme using only three pilot selectors. In today’s programming work there is always a
choice of goals: we can work to save storage space, or execution time, or
compile time. In those halcyon days
there was only one goal: to save storage space. You might reflect on how many of your current
programs would fit into 88 words of data storage.
And there was another thing: what you might call built-in intermittents. When storing a result on the CPC, it was necessary to allow two instructions to intervene before the stored number could be recalled. That seems reasonable, but if this rule were violated, about one-third of the time the thing would work anyway. Thus, with improper programming it was possible to have the test cases check out perfectly but the production runs would be wrong. According to McDougal’s Law, this is exactly what happened.
So computing in those days had a different flavor about it. It seems ages since we dealt with morning sickness, for example (the perverse gadget that wouldn’t work at all at 0800 would magically cure itself just as the maintenance man arrived at 1000). We get a lot more done today (one 7094 does more in a day than all the CPC’s ever did), but some of the fun is gone. A dozen years from now, will we be writing about the 7094 in the same nostalgic way?
From Datamation (a defunct computer magazine), June 1964