Technical Travels and Travails

Image: Cray-1 Super Computer

Higher Math

About Me

On October 15 the United States Mint released designs for next year’s American Innovation coins. One of the coins honours the Cray-1 supercomputer, developed in Wisconsin.

In 1972 Seymour Cray left Control Data Corporation to form Cray Research. The site for the new lab in Chippewa Falls, Wisconsin, was quite close to the CDC lab he left. After years of research and design the Cray-1 was announced in 1975.

The Cray-1 used a large number of ECL gates arranged in a “C” shape to minimize path lengths for the wiring. The processor units were surrounded by power supplies and refrigeration located under the padded “seats”. The main unit was about nine feet in diameter and seven feet high (263 x 196 cm) and weighed in at five tons (4800 kg).

By far the most popular programming language on the Cray-1 was a version of Fortran that supported automatic vectorization. Other languages were available, however.

In the late 1980s a Cray-1 XMP was installed on the University of Toronto Computing Services raised floor. The weight and size of the machine was such that a hole had to be made in the roof so that a crane could lower it into place. The system included a team of support people who informed us that we were the fifth Canadian installation that they supported. They also had charge of an “unspecified” location in Ottawa. We assumed that this was on Olgilvie Road.

I supported the administrative computing at the time so I did not have direct involvement with the users. I did, however, get involved with billing for services. The Cray-1 collected suitable metrics for for the purpose, but they were kept in Cray’s unique floating point format which our IBM mainframe-based billing system couldn’t interpret. We ended up writing a program on the Cray which would read the counters and write out a text file for further processing. We decided not to use Fortran for this purpose given its limited I/O flexibility and instead used Pascal.

It is unlikely that our billing program benefited much from vectorization.

I graduated from Engineering at a time when slide rules were just being replaced by calculators as being acceptable for exam use. I discovered that I had a talent for using computers to solve engineering problems, which led to a career.

I have worked in the IT business for many decades, spanning mainframes to mobile. I have seen a lot of technologies come and go with many reappearing in a different guise. Having witnessed many Silver Bullet ideas fail to fire, I remain optimistic.

LinkedIn