Fifty years later, IBM's inventors celebrate the 'Stretch'

Judged a failure at the time, Big Blue's 7030 supercomputer left a rich, albeit indirect legacy for the rest of the computer industry.

Charles Cooper Former Executive Editor / News
Charles Cooper was an executive editor at CNET News. He has covered technology and business for more than 25 years, working at CBSNews.com, the Associated Press, Computer & Software News, Computer Shopper, PC Week, and ZDNet.
Charles Cooper
3 min read

Left to right: Stretch collaborators Fran Allen, Fred Brooks, and Harwood Kolsky, and The New York Times' Steve Lohr. Computer History Museum
So how does it feel to be associated with one of Big Blue's biggest failures?

"All depends on your perspective," recalled an amused Fran Allen, not at all regretting her participation in a now-storied mid-1950s supercomputer project popularly known as "Stretch."

Even though IBM only built nine of the machines, Stretch left behind a legacy that remains a source of pride to the participants who were present at its creation.

"A lot of what went into that effort was later helpful to the rest of the industry," Allen said with the sort of understatement you'd expect from a former winner of the prestigious Turing Award. Fact is that Allen and the 300-some people who collaborated on Stretch invented many of the concepts that later became standard computer technologies. The short list includes multiprogramming, pipelining, memory protection, memory interleaving, and the eight-bit byte.

Stretch: announced April 1960 and withdrawn in 1961. IBM

Many members of that original team, now grayer and more slow-of-gait than they were during the Eisenhower administration, filled an auditorium Thursday night at the Computer History Museum to reminisce and consider the legacy they bequeathed. Fred Brooks, who was a system planner for Stretch, and Harwood Kolsky, who worked on product planning, later joined Allen on stage for a panel discussion moderated by The New York Times reporter Steve Lohr.

In January 1956, work on the Stretch project formally got underway with the goal of building a supercomputer to replace IBM's 704 supercomputer. The resulting product, called the 7030, as the Stretch was officially known, could perform 100 billion computations a day and handle half a million instructions per second.

"We were about 300 people working in Poughkeepsie (New York)," said Kolsky. "Individual teams met frequently. That's why it's hard to tell who invented what. Generally, morale was high. You wouldn't know it by looking up here, but it was a young person's group...there were only two people over 40. Most members of the team were in their 20s and 30s."

But they were in for a shock. IBM's then-CEO, Thomas Watson, Jr. judged the 7030 to be a failure. Even though the machine was about 30 to 40 times faster than other systems, IBM won a bid submitted to Los Alamos Scientific Laboratory on its pledge to build a supercomputer that was at least 100 times faster than the 704.

When Stretch came along, IBM, which controlled about 70 percent of the computer market and about 90 percent of the punch-card business, was already fending off charges it exerted monopoly control. Brooks said that when Watson ordered the original price cut to $10 million, "that put it at under cost and violated antitrust...antitrust was a fact of everyday life in all our thinking."

In fact, IBM had signed off on a consent decree with the Justice Department in 1956. The company eventually shipped nine systems to customers around the world but then closed the production line forever.

Kolsky recalled the initial reaction to Watson's decision, saying that the project's engineers thought it served as a potential death knell to future supercomputer development.

The first system was developed for the Los Alamos Scientific Laboratory under contract to the Atomic Energy Commission Computer History Museum

Of course, it was anything but. In fact, the lessons learned from Stretch paved the way for the subsequent development of IBM's System/360, which turned out to be a smash success. Meanwhile, the innovations invented for Stretch subsequently entered the wider world of mainstream computing.

With the exception of industry cognoscenti--and the relative handful of folks responsible for engineering and managing the project--Stretch remains a footnote for most people. But maybe that's starting to change. To underscore the moment, IBM flew in one of its up-and-comers, senior VP of Development and Manufacturing Rod Adkins, to introduce the panel. Earlier in the evening, I sat down for a conversation with Adkins, who placed Stretch in its historical context.

"It's a pretty good model of a highly ambitious program that, at the time, was considered having not met its objectives," he said. "But when you look at some of the things that came out of that effort and how it's influenced the computer industry today, (Stretch) has had a profound, indirect, benefit to this industry."

Computer historians should also note the following: Stretch remained the most powerful computer in the world until 1964. Some failure.