Period 1978 - 1983: Simulations expand

The Control Data CYBER 74 system

Early 1978, it became clear that the CDC6400 was so overloaded that expansion of the CPU capacity, of the disks and the memory had to take place. The replacement of the CDC 6400 by a CDC CYBER 74 was scheduled. The largest applications were simulations for tanks and the "Wargame"-application continuously increased as new models of military equipment were developed.

CYBER 74 hardware

The CDC CYBER 74 system was a modern version of the Control Data6600 . The system had a clock cycle of 100 nanoseconden (10 Mhz), a ferromagnetic memory of 131 Kwords of 60 bits (no parity) and ten (afterwards 14) peripheral processors (1 msec major cycle). The system required 63 KWatt per hour. The heat was taken away mostly by an impressive water cooling system and required an air conditioning as well. To keep the system working, it required a three-hours preventive maintanance weekly!

Installation of the CYBER 74: re-organization of the electric power facilities

Installation of the CYBER 74; three of four 'bays' are already connected.
The so-called "christmas tree", a pole between the bays that took the weight
of all cross-bay wiring is not yet installed.

For the users a number of restrictions were built into the NOS/BE operating system. For one interactive process-step, a maximum of ten seconds calculation (CPU)-time was allowed. Batch jobs taking more CPU-time than 7.5 minutes were only processed at night and in week-ends, with exception of "tape jobs" (you could trick the system!). At the same time, batch jobs could not use more than 140000B of memory during daytime. At night this limit was 200000B words (369 KB resp. 429 KB). Interactive processing was limited to 70000B words (215 KB).

Installation problems

The installation of the CYBER 74 in August 1978 took three weeks. At first the Control Data 6400 was removed. The computer room was expanded with one Laboratory room (roughly 5*4 meters). During this replacement period, the Laboratory had no computing facilities at all! (yes, we could live with that!)
Because each of the four "legs" (bay's) of the system weighed about 2000 kilos and after installation the weight also would be increased by the cooling water running through the bays, the architect of the building had to calculate whether the floors of the building would have sufficient bearing-strength. This gave no problems. However special precautions had taken to pull up the system parts along the slanting slope of the heigthened computer room: thick balks against the outside of the outer wall to connect the tackles. Not to disturb the daily Laboratory internal transports, the decision was made to do the vertical transport of the heavy system parts on Sunday. Previously was investigated if that weight of the parts could be elevated by the elevator. During these preparations one thing was forgotten. The elevator overload device had to be shunted from normal operation into heavy load operation. On the Sunday that the first bay had to hoisted, no elevator service man was ordered to be present. The only solution was to hoist up the parts by turning the cabledrum of the elevator many times by manual power…

Start up problems

Because the new tape controller also needed new 'controlware', the so called "coldstart" controlware program had to be read by the punchcard reader operated by a special "deadstart program" (set by switches) into the tape controller. The problem was that the Laboratory had no punchcard puncher and that the coldstart program was delivered on a magnetic tape. A chicken-or-egg problem. Very creative work by an engineer with help of the system programmers, lead to a bootstrap process in which a coldstart deck was manifactured ("ABC"-program) to wake up the new CYBER.

Radar radiation

At the same time as the installation of the CYBER 74, the lay-out of the computer room was changed. An air circulation separation was installed between the "dusty" print and card reading areas in the room and the (dust sensitive) magnetic tape handling area and the harddisk devices. This caused the number of disk errors to decrease spectacularly. Hardware problems were daily monitored by means of a system log (CERFILE) reporting tool (HPA). However, the first month with the new lay-out showed a very large number of read errors on the magnetic-tape units. Analysis demonstrated that the problems only occured when processing magnetic-tapes of the users. The processing of the back-up tapes showed the usual couple of read/write errors as we were used to before the replacement of the system. Silly was, the problems were not originating a particular magnetic tape-unit and more often used magnetic tapes showed at one time the "usual" errors and the next time several thousands of read errors.

CYBER 74 console and magnetic-tape units
The radar sensitive magnetic tape units and the CYBER 74 console.
From behind the air cooling, that often leaked cooling water.

Still digging deeper, it was demonstrated that the errors - if they occured - showed a frequency of 0.1 Hz. The "Bedrijfstechnische Stafafdeling" inspected the electricity grid. That seemed to be OK. Meanwhile, systems programming expanded the operator console display program with a facility to display tape wrinting/reading errors on-line on the console screen.

During several weeks the 10-seconds magnetic tape problem did not occur. At the moment we had the feeling that the "10 seconds ghost" was gone, the problem appeared on a Friday afternoon. Every ten seconds, a tape read error occured. Within the shortest time, the facility management department of the laboratory was warned and more than ten people looked at the phenomenon at the console. A second magnetic-tape unit was started. At that time, exactly every ten seconds two messages were logged. One of the people present became bored and took a look outside through the window. He saw the usual skyline view with the radars of STC, the neighbours of the Laboratory. And there, a radar turned around with a speed of ten seconds exactly. One phone call to STC requesting to turn off the radar for some time resulted in an angry researcher: his radar could not jam computers! After some discussion, he realised himself that the sidelob of his radar beam radiated on the heads of our magnetic-tape units (that were turned 90 degrees in the new lay-out; before that their backs were directed towards the radar). Affixing a piece of grounded chicken wire in all windows solved the problems once and for all.

Anyway, also at the LEOK-location the same type of problem occured after the movement of some radar equipment. According to the radar experts, a line of trees reflected the radar radiation into the computer location. After receiving the tree cutting permission from the city, the complete line of trees was chopped. It did not solve the problem!
It turned out that the radar radiation was directly transmitted through the roof! At this location, a simple Faraday cage solved the computer problems.

MuseumWaalsdorp@tno.nl