Dinosaur Computers
from Gerhard Holtkamp, 06. August 2009, 19:28
Early computers and dinosaurs had something in common - they were physically large and became extinct. Here is a personal account...
One of my first activities after starting university was to enroll in a computer course. This was long before the advent of PCs. Back then computers were very expensive. Large companies, some government installations and universities were about the only places where you could find them.
Soon the big moment arrived for me to run my very first computer program. Just entering a few numbers, performing a simple calculation and outputting the result. The usual Hello World program that everybody likes to start with. What was unusual out of todays perspective was the type of computer I used: A Zuse Z22.
Developed in the 1950s the Z22 was a first generation computer which used 600 vacuum tubes for its logic circuits. Tipping the scale at one ton it took up a room all for itself. Much of the 3500 Watt of electricity needed to run this machine would turn into heat which necessitated a good ventilation plus some water pipes. The vacuum tubes wouldn't last too long and often needed replacement causing a lot of down time.
The memory of the Z22 was divided in words of 38 bit length. The magnetic core memory (somewhat equivalent to what we call RAM nowadays) had just 14 words. There was also what we proudly called a mass storage device: A magnetic drum able to store 8192 words. Not even the most resource optimized Linux distribution would be able to run on something like this.
Luckily the Z22 didn't need an operating system. There was a console with lit up buttons that an operator had to press in order to initiate certain actions. A special teletype writer served as combined input/output device. You could enter something via the keyboard but usually programs and data were written on rolls of (5 channel) punched paper tape which was accepted by this teletype writer. Output was usually written on paper but you could also have it punched on tape.
The lowest bits of the main arithmetic register were connected to a loudspeaker. This proved to be most useful. By listening you could figure out what the computer was doing. Calculating square roots (which took half a second on the Z22) produced a certain sound which was very different from what you would hear if the program got stuck in an endless loop.
Without an operating system there was no multi-tasking. Only one job could run at any one time. For programming we had a choice between assembler language or ALGOL. ALGOL (together with FORTRAN) was the first higher level computer language. It had the basic elements for mathematical algorithms needed in science and engineering.
As the ALGOL compiler needed a certain amount of our very scarce memory it would allow a program to also use the compiler memory space if needed. Of course if you ran a program like that you would have to reload the compiler afterwards from a large roll of punched paper tape which took some 10 minutes or so.
We prepared our programs off line at standard teletype writers which were quite common in those days at news agencies. You could see what you were typing on paper while the paper tape was being punched. Then as now you made errors in typing and also in programming. To fix these errors there were several techniques. We were all familiar with the 5 channel code used and manually punching an extra hole or two with a blunt pin could change a wrong character into a right one (or an inert one). You could also copy pieces of a paper tape and then continue writing. But if all that didn't do you always had a pair of scissors and some glue nearby for the original "cut" and "paste"!
With memory at a premium we sometimes held competitions of given a certain problem who could write the program with the least amount of memory. This often resulted in some highly convoluted code which was difficult to understand and to maintain. Very labor intensive but it explains why we were able to do anything at all with such a computer while today's generation is used to an almost limitless supply of memory.
I still consider it my greatest achievement at the time to have come up with a three line ALGOL program that completely wiped out all the memory. (This had been unintentional I should add and was due to a compiler error!) After that you had to start from scratch. The poor operator first had to manually enter a list of words by pressing the respective buttons on the keyboard bit by bit (yes that's right - one bit at a time!). This was kind of a primitive BIOS as we would call it today. After that you could finally use rolls of punch tape to properly boot the system and load the assembler and compiler. During the half hour or so which this took all computing activities at that university came to a standstill.
Well not quite. We had a second computer, a Zuse Z23. As you might guess the Z23 was the follow on model to the Z22. It was about the same size and weight but used transistors instead of vacuum tubes. Its working memory was 248 words and it ran about three or four times faster than the Z22. For more efficient output it even featured a large line printer (only capital letters in those days!). Programs were usually compatible with either machine so you could use whichever was available.
Sometime during my second year at the university we got a new computer. Larger, faster and much more memory. Our main input now advanced to punch cards and we also had magnetic tapes available. It still used individually mounted transistors but had already a very rudimentary operating system (yet still only one job at a time).
While our Z23 was transferred to another place for a brief stint before retirement the old Z22 had by now become more valuable as an antique object and was eagerly taken up by a museum. We must have been one of the last places to ever use a machine like that. While we young folks were very excited about our new computer I could see a few tears in the eyes of the director of our computer department when the Z22 was taken away. An era in computing history had just ended.
It was about 25 years later in June 1993. The D-2 Spacelab mission had taken place a month ago and this was a time of intense debriefing and partying by the participants. One morning the astronauts and operations team (of which I was a part) were invited to a reception and press event at the German ministry of science and technology.
We were brought into a very large room half filled with chairs for us and otherwise almost empty. While waiting for the minister to arrive and give his speech ("You have done a brilliant job...", "The country can be proud...", "We don't have enough money to let you continue like that...") I realized that there was something in one of the corners of the unused part of that room. It happened to be a Z23. Why it was there I don't know as there were no other museum pieces present. Maybe some temporary storage (many government facilities were about to move to Berlin).
I went over to the familiar Z23 console. It was odd seeing it at this place and time. Having just worked with some of the most state-of-the-art equipment during the Spacelab mission I could only wonder.
We had come a long way in a short time...
One of my first activities after starting university was to enroll in a computer course. This was long before the advent of PCs. Back then computers were very expensive. Large companies, some government installations and universities were about the only places where you could find them.
Soon the big moment arrived for me to run my very first computer program. Just entering a few numbers, performing a simple calculation and outputting the result. The usual Hello World program that everybody likes to start with. What was unusual out of todays perspective was the type of computer I used: A Zuse Z22.
Developed in the 1950s the Z22 was a first generation computer which used 600 vacuum tubes for its logic circuits. Tipping the scale at one ton it took up a room all for itself. Much of the 3500 Watt of electricity needed to run this machine would turn into heat which necessitated a good ventilation plus some water pipes. The vacuum tubes wouldn't last too long and often needed replacement causing a lot of down time.
The memory of the Z22 was divided in words of 38 bit length. The magnetic core memory (somewhat equivalent to what we call RAM nowadays) had just 14 words. There was also what we proudly called a mass storage device: A magnetic drum able to store 8192 words. Not even the most resource optimized Linux distribution would be able to run on something like this.
Luckily the Z22 didn't need an operating system. There was a console with lit up buttons that an operator had to press in order to initiate certain actions. A special teletype writer served as combined input/output device. You could enter something via the keyboard but usually programs and data were written on rolls of (5 channel) punched paper tape which was accepted by this teletype writer. Output was usually written on paper but you could also have it punched on tape.
The lowest bits of the main arithmetic register were connected to a loudspeaker. This proved to be most useful. By listening you could figure out what the computer was doing. Calculating square roots (which took half a second on the Z22) produced a certain sound which was very different from what you would hear if the program got stuck in an endless loop.
Without an operating system there was no multi-tasking. Only one job could run at any one time. For programming we had a choice between assembler language or ALGOL. ALGOL (together with FORTRAN) was the first higher level computer language. It had the basic elements for mathematical algorithms needed in science and engineering.
As the ALGOL compiler needed a certain amount of our very scarce memory it would allow a program to also use the compiler memory space if needed. Of course if you ran a program like that you would have to reload the compiler afterwards from a large roll of punched paper tape which took some 10 minutes or so.
We prepared our programs off line at standard teletype writers which were quite common in those days at news agencies. You could see what you were typing on paper while the paper tape was being punched. Then as now you made errors in typing and also in programming. To fix these errors there were several techniques. We were all familiar with the 5 channel code used and manually punching an extra hole or two with a blunt pin could change a wrong character into a right one (or an inert one). You could also copy pieces of a paper tape and then continue writing. But if all that didn't do you always had a pair of scissors and some glue nearby for the original "cut" and "paste"!
With memory at a premium we sometimes held competitions of given a certain problem who could write the program with the least amount of memory. This often resulted in some highly convoluted code which was difficult to understand and to maintain. Very labor intensive but it explains why we were able to do anything at all with such a computer while today's generation is used to an almost limitless supply of memory.
I still consider it my greatest achievement at the time to have come up with a three line ALGOL program that completely wiped out all the memory. (This had been unintentional I should add and was due to a compiler error!) After that you had to start from scratch. The poor operator first had to manually enter a list of words by pressing the respective buttons on the keyboard bit by bit (yes that's right - one bit at a time!). This was kind of a primitive BIOS as we would call it today. After that you could finally use rolls of punch tape to properly boot the system and load the assembler and compiler. During the half hour or so which this took all computing activities at that university came to a standstill.
Well not quite. We had a second computer, a Zuse Z23. As you might guess the Z23 was the follow on model to the Z22. It was about the same size and weight but used transistors instead of vacuum tubes. Its working memory was 248 words and it ran about three or four times faster than the Z22. For more efficient output it even featured a large line printer (only capital letters in those days!). Programs were usually compatible with either machine so you could use whichever was available.
Sometime during my second year at the university we got a new computer. Larger, faster and much more memory. Our main input now advanced to punch cards and we also had magnetic tapes available. It still used individually mounted transistors but had already a very rudimentary operating system (yet still only one job at a time).
While our Z23 was transferred to another place for a brief stint before retirement the old Z22 had by now become more valuable as an antique object and was eagerly taken up by a museum. We must have been one of the last places to ever use a machine like that. While we young folks were very excited about our new computer I could see a few tears in the eyes of the director of our computer department when the Z22 was taken away. An era in computing history had just ended.
It was about 25 years later in June 1993. The D-2 Spacelab mission had taken place a month ago and this was a time of intense debriefing and partying by the participants. One morning the astronauts and operations team (of which I was a part) were invited to a reception and press event at the German ministry of science and technology.
We were brought into a very large room half filled with chairs for us and otherwise almost empty. While waiting for the minister to arrive and give his speech ("You have done a brilliant job...", "The country can be proud...", "We don't have enough money to let you continue like that...") I realized that there was something in one of the corners of the unused part of that room. It happened to be a Z23. Why it was there I don't know as there were no other museum pieces present. Maybe some temporary storage (many government facilities were about to move to Berlin).
I went over to the familiar Z23 console. It was odd seeing it at this place and time. Having just worked with some of the most state-of-the-art equipment during the Spacelab mission I could only wonder.
We had come a long way in a short time...
Reading your post sent me down memory lane.
I remember that as a young student, just 19 years of age, I watched a crane removing some heavy machinery from the first floor of the main building of my university. The workmen told me that they were taking out the old computers and the punch card readers. All of these devices were too unwieldy to be moved down the stairs. Luckily the windows in this 18th century building were large enough. Very perspicacious of the architect who lived 200 years ago I must say.
Those computers were replaced by a modern (by 1981 standards) mainframe computer, which constituted my first - and up to the time I purchased my first C64, only - hands-on contact with the world of electronic information technology.
Looking back today and talking to the youngsters who have just left university, it seems inconceivable to me that there is a whole generation of budding computer scientists who have never mounted a data tape, never spent hours bathed in the eery glow of a monochrome, character-only screen and cannot imagine the time when computer experts associated "mouse" with a small, furry animal that would build its nest somewhere in the innards of a computer and gnaw important cables.
You mentioned another dinosaur, Fortran. However, unlike the long-gone hardware that spawned it, Fortran is still among us and very well alive, a living testimony to the fact that a program once used to be a stack of cards with holes punched into it. Although ADA and especially C++ are making some inroads, I see no sign that the era of the reign of king Fortran is coming to an end.
Some things go full cycle. When I started out on my first job, photocopy machines were expensive and rare, and one had to walk to the opposite end of the bulding if one wanted to make a photocopy. With time, they got cheaper and better, and there was one on every floor, so one did not have to walk nearly as far.
Now, it seems that the newest generation of these machines can do everything, send faxes, send e-mails - and of course make simple photocopies as well - but the downside is that they are again very complex and expensive and therefore fewer of these sleek machines are installed - so once again, I find myself having to walk to the opposite end of the building just to copy one page.
Plus ca change, plus c'est la meme chose.
(Your quip on the speech of that minister did not pass unnoticed. However, I must add that if that was who I think it was, then he was the one German science and technology minister in recent history to make an honest effort to promote space activities rather than just doing the bare minimum)