Next / Previous / Shipman's Home Sweet Homepage / Site map

The computer torturers

Tech's first computer, an IBM System/360 model 40, arrived in the summer of 1966, about a week after I got here. The grounds crew got to unload it, and I still can't believe that the machine survived this process.

The processor cabinet weighed about a ton and was six feet long. It was sitting in the truck with the narrow end toward the back. The forklift operator roared up to the back of the truck and stuck the forks under the box. The forks were not quite half as long as the box. It never occurred to the operator that he couldn't pick up a long box with those short forks.

He got the forks stuck in at more or less the correct angle and started to lift. The forks went up about two inches, and then the processor tipped and fell off the forks---just a short fall, fortunately.

The grounds crew people scratched their heads about this for a few minutes, and then one of them got an idea. He got a length of webbing (cloth furniture strapping), ran it around the box horizontally, and fastened it. Then he gave the signal to start lifting again. It didn't look to me like that webbing would be strong enough to hold the box, but I didn't say anything because I assumed the grounds crew knew what they were doing, and I was a brand new freshman so I didn't think they would listen to me even if I did speak up.

About this time, Dr. Tom Nartker, the computer center director, came out of the building to watch the unloading. When he saw the crew trying to lift the box with only a piece of webbing holding it to the forklift, he came unglued and started shouting at the forklift operator to please put it back down---gently.

Dr. Nartker knew that computers were delicate and they wouldn't survive rough handling. He convinced the grounds crew to try to fasten the box to the forklift a bit more solidly.

Their next idea was to get some heavy logging chain. They attached one end of the chain to a beam underneath the forks and then ran it under the processor, around the back end, and over the top. They stuffed some furniture pads under the chain so the cabinet wouldn't get scratched. The free end of the chain was attached to a large ring at the top of the forklift. This arrangement satisfied Dr. Nartker, so they started trying to lift the box again.

Unfortunately, the bottom of the chain was attached to the moving part of the forklift, while the big ring at the top didn't move with the forks, so when the forks went up, the chain got tighter and tighter. The forklift was making groaning noises and the operator couldn't figure out why. Then a two-inch-square frame member on the back of the processor cabinet buckled under the tension.

The IBM field engineers had asked us to wait until they got there before trying to unload the machine, and we probably should have waited---they definitely had some good ideas. They arrived while Nartker and the grounds crew were arguing about what to do next. As it turned out, there were some casters installed inside the corners of the cabinet, and all it took was a few turns with a wrench to lower the casters. Then it was a simple matter to roll the cabinet around so that the long side of the cabinet was toward the back of the truck, and the forks were long enough to pick up the cabinet across its width.

The forklift operator was proud of his ability to whip the forklift around in a hurry, and it took a lot of shouting and threatening to convince him that he should take it slow with this quarter-million-dollar machine.

While we watched the machine inching out of the truck and being lowered the eight feet to the ground, the IBM field engineer told us an amusing story about another system that had been sold to a company in Dallas---a 360/91, which cost several million dollars and was the supercomputer of its day.

The people who picked up this machine at the Dallas airport were very conscientious. They loaded it very gently into an air-ride van and drove the van at ten miles an hour through back streets to get to the plant. They eased it off the van and into the elevator and took it up to the computer center floor. When they opened up the cabinet, inside there was a large pile of broken and mangled parts. Apparently someone at the airport had unloaded the crate by pushing it out of an airplane fourteen feet above the runway.

Fortunately, our machine had not been badly mangled by the delivery process, although we had to put up with a few snide remarks about the bent frame.

This computer system did not have a lot of the things that modern computer users take for granted. It had no disc storage, just four tape drives each about the size of a telephone booth. Although most computers in those days had a typewriter-like operator's console, this machine had none. Input was through punched cards only, and any messages to the operator would come out on the line printer. The operating system was called TOS (Tape Operating System), an incredibly ancient and backwards technology even for its day.

The only possible operation was to compile and run a FORTRAN program. This required that the operator mount four tapes, one on each tape drive. One of the tapes was the ``system tape'' containing the operating system, FORTRAN compiler, and loader. The other three tape drives held scratch tapes that were used for intermediate results during compilation. The operator loaded a deck of punch cards into the card reader, rebooted the operating system from tape, and watched the tapes spin for a while.

Dr. Alan Miller (still at Tech) was one of the early users of this system; he can testify about operation in those days. The operator would usually run a job as soon as it was submitted, and then he would take some time to discuss the results with the user. It was several months before there were enough users that people had to wait before their job was run.

The computer center was located in the second floor of the tower in R&DD (now Workman Center). Computers in those days generated a lot of heat, so we had three large refrigerative air conditioners hanging out the windows, and they should have been sufficient to keep things cool. However, anyone who has ever used a refrigerative air conditioner in humid weather knows that they tend to accumulate ice in the cooling coils and lose efficiency.

In those days, computers had to be kept quite cool---under 65 degrees Fahrenheit. Typically one of the coolers would ice up, and the room would start to get warm. So the operator would crank all three of the coolers to maximum cooling, which would make them ice up faster. Pretty soon the room would get too hot even for humans, and usually the machine would fail.

The front panel had a lot of blinking amber lights on it for tracking normal operation. Experienced operators could watch the lights flash and get a pretty good idea of what the machine was doing. There was a double row of red indicator lights on the upper left corner of the panel that were not supposed to go on unless something bad were happening. So usually when the room warmed up, one or more red lights would go on, and we would shut the processor down and call IBM in Albuquerque.

Of course, by the time the field engineers drove down from Albuquerque, the room would be nice and cool again. They used to fume about how often our system failed; they never suspected how badly we tortured their poor machine. Once we borrowed a thermometer from the instrument room and put it on top of the cabinet where the cooling air was coming out of the memory modules. We were rather appalled to see that it read 140 while the room temperature was only about 85 or so.

When we ordered the system, we didn't have enough money for the normal card reader and printer devices that most people used on System/360 machines. IBM had a special line of cheap peripherals for customers like us: the 1442N1 card reader and 1443 line printer. They broke even more often than the processor did.

After we had had the system for about a year, the field engineers informed us that our card reader and printer failed more often than any other units of these same models anywhere in the world. To study these failures in more detail, they shipped us brand new replacements for both units, and took the old ones back to the factory to see if they could find out why they had failed so much. Of course we never told them about the cooling problems; it would have invalidated the warranty.

The fact that lightning often hit the tower only a few floors above the computer center probably also had something to do with the failure rate.

Years later we found out that it wasn't entirely our fault that the printer broke down so often. The model 1443 printer was originally designed as a backup logging device for system consoles; it was supposed to print a duplicate listing of all the messages that came out on the console. It wasn't designed for heavy, continuous operation. IBM didn't admit it, but the 1443 was rated for a 25% duty cycle. Of course, we were using it eighteen hours a day!

Next: An emergency landing
See also: Shipman's original writings
Previous: An occultation on the Rio Grande
Site map
John W. Shipman,
Last updated: 1997/01/25 20:50:09