"A Walk Through 'Visible Storage'", section 3 of 6, by LEN SHUSTEK
From "CORE 2.3", a publication of The Computer History Museum.

Forward, Backword, Table of Contents, Docent Page


BUILDING SUPERCOMPUTERS

Many of the artifacts in the collection demonstrate technological or commercial failures, and studying these is one of the best ways to learn from history. The "STRETCH,"


Photo by Michael Dubinsky IBM's attempt in the late 1950s to build a supercomputer dramatically better than anything that had come before, was a commercial failure because it was too expensive and not fast enough. But it pioneered amazing technology that later surfaced in other computers over the next 20 years. Due to its commercial failure project engineer Red Dunwell was considered a persona non grata by T.J. Watson for many years, but later was lauded by Watson when STRETCH's numerous innovations had become apparent.

Although IBM was very successful in providing computers for the military and for ordinary businesses, from STRETCH onward, and for the next several decades, it struggled with building the very fastest scientific computers. In 1965, a small company in Minnesota introduced the CDC 6600,


Photo by Jessica Huynh
which tweaked IBM's nose by being the fastest computer in the world for many years. An angry T.J. Watson blasted his staff with this memo: "Last week, Control Data ...announced the 6600 system. I understand that in the laboratory ...there are only 34 people including the janitor. Of these, 14 are engineers and 4 are programmers .... Contrasting this modest effort with our vast development activities, I fail to understand why we have lost our industry leadership position by letting someone else offer the world's most powerful computer."

Sometimes, Mr. Watson, bigger isn't better.

Part of CDC's advantage over IBM was its smallness, but part was the remarkable genius of its principal designer, Seymour Cray. He got his start designing computers for the military, like this Univac NTDS


Photo by Jessica Huynh
computer used on board a battleship and built like a tank. In general, the military's influence in the early development of computers was huge and the industry would not have developed as quickly without it.

Seymour Cray


had a long and distinguished career based on repeatedly designing the world's fastest computers until his untimely death in a car accident in 1997. This Cray-1

Photo by Jessica Huynh
from 1976, sometimes called the "world's most expensive loveseat," is perhaps the most famous example.

The physical design of fast supercomputers presents two important problems: keeping the circuitry close together so that delays caused by wiring are minimized, and getting the heat out so that circuits don't overheat. In speaking about this machine at the time, Cray was as proud of the plumbing that kept it cool as the electronics that did the computing, and would talk at length about his patents for copper tube extrusions into the aluminum cooling columns.

Cray's next machine, uncreatively called the "Cray-2,"


Photo by Michael Dubinsky
solved the cooling/plumbing problem another way: the boards themselves were swimming in a non-conducting liquid called "Fluorinert," a blood plasma substitute used in surgery that happens to have the right thermal, mechanical, and electrical properties. Changing out a defective board within the 30-minute "mean time to repair" requirement was a challenge, though, since the Fluorinert had to be pumped into a holding tank, the board replaced, and the liquid pumped back.

Other people solved the cooling problem for supercomputers in other ways. The "ETA-10,"


Photo by Jessica Huynh
created by engineers who, like Cray, had left CDC, contained circuit boards that were immersed in a vat of liquid nitrogen.



MAKING MAINFRAMES

In the meantime, IBM was doing a booming business selling mid-sized computers for both business and scientific purposes. But by the early 1960s it had a looming crisis: it was building too many different kinds of computers. Each used different technology, software, engineers, salesmen, and support technicians. To consolidate behind a single uniform product line that could do both scientific and business computing at both small and large scale, Watson put a 28-yearold untested manager by the name of Fred Brooks in charge of a "you bet your company" project that would obsolete their entire product line. It was a remarkably bold move for a 60-year-old prosperous company and it could have been a colossal disaster, but the result was wild success: IBM dominated the mainframe computer industry for 20 years with the System/360


Photo by Jessica Huynh
that was introduced in 1964.

Within the 360 family, IBM did finally manage to build a supercomputer that could compete with the CDC 6600. And this IBM 360/91


Photo by Jessica Huynh
was perhaps the pinnacle of the "lights and switches" front console design, although even by then most of the operation and fault diagnosis of computers was being done electronically. The end of the flashing lights and whirring tape drives since then has made computers more efficient but much less photogenic.


go to top
Forward, Backword, Table of Contents, Docent Page