Dai Edwards: the first stored program computer
Now at that time there was computing on the ENIAC in the States, which was a very large machine, 18,000 valves. Its memory was on punch cards, so obviously the access to it was very slow and things. And whilst that could provide the data in and the data out, the programme had to be set up by setting switches on the various units and connecting cables. And depending on the complexity of the problem and getting the connections all right, this could take days if it was simple, but obviously up to weeks if it was much more difficult. So the idea of the cathode ray tube store was when your program were right and you introduced them, you just had to feed the information into the memory, the computer would obey it. Shortly afterwards you could replace that program with another program, and it would run that. And of course the time to do that was relatively short, certainly compared with a day [laughs]. It came as a bit of a surprise to us to find out what was going on, because earlier in the year when the machine had actually worked, in June, we were undergraduates in the department and nothing had ever filtered through to us. So the first thing was very surprised. The next thing was when we saw the equipment. It obviously looked something like the thing that’s at the back of us, the machine at the back of us now which is a replica, but of course I think the actual machine was not quite as clean and the wires were not quite as tidy either. But the thing that impressed me was that all the valves seems to be put in from underneath, and I thought there was a danger that these might actually drop out, but that never actually seemed to happen. But we were really enthused because we felt there was great potential in developing this into a user system and there seemed to be problems around which people needed to have this sort of machine for. We already knew about areas where people were using teams, using mechanical calculators, fifty mechanical calculators, to try and solve their problems, and they were desperate to get something which was more capable. The Baby was built by FC Williams and Tom Kilburn and Geoff Tootill. They built it in conjunction with technicians in the university who actually built the chassis, cut the metal, you know, obtained the racks and all that sort of thing. The CRT, just like this one, has been used, as it were, originally for producing an instrument, like an oscilloscope, but now we’re talking about using it as a memory in a computer. And the idea is that you can store on this screen, which you can see the white screen, you can store in separate places on this screen, discreet places, you can store different charges. And those charges can be coupled to a metal plate which covers the screen, so when they change that signal is coupled to the metal plate and you amplify it and you can detect what they are. And you can detect a nought and a one. And you can in fact display that on a raster where there’s a picture showing a thirty-two by thirty-two display of distinct areas where information is stored. Each of those can be a one or a nought. What the Baby machine here was, you know, originally built for, was to test that it not only worked statically to show steady pictures, but it could be used with all that information being changed at electronic speeds. And that was the reason for building the Baby computer, it was a test bed for that purpose. I mentioned the ENIAC already that it was doing digital computing, but it was a fixed programme machine. The Baby demonstrated the fact that you could now make a programmable computer.
Related Audio Clips
The following clips are short extracts from an in-depth interview.
To listen to the full interview visit http://sounds.bl.uk