The IBM 7030, aka Stretch
IBM enters the age of supercomputing
An engineer at the control console for an IBM 7030

The IBM 7030, introduced in 1960, represented multiple breakthroughs in computer technology. It was IBM’s first supercomputer, ranking as the fastest in the world for three years after its debut. Rather than relying on bulky and often unreliable vacuum tubes, it used transistors. And its advanced random access disk drives provided unrivaled data storage and retrieval speed. The 7030 stretched the limits of computer design to such an extent that it soon became known by the nickname IBM had given it during its development — Stretch.

Stretch was created to meet the rigorous computing requirements of government laboratories engaged in nuclear weapons research. Nuclear scientists often needed to conduct more than 100 billion arithmetical operations to evaluate a weapon’s design. Previously, the world’s fastest computers needed six months to solve one of these problems. Stretch could complete it in about a day. The first Stretch was delivered to Los Alamos National Laboratory in 1961, and a second customized version, the IBM 7950 Harvest, went to the National Security Agency in 1962 for use in cryptanalysis. Government laboratories in England and France later purchased Stretch systems to perform weapons research.

Stretch achieved its high speeds through the principle of simultaneous operation. The processing unit was organized like an assembly line. As each part completed a task, it passed the work to another machine element and then started a subsequent task. This enabled the main arithmetic unit to be almost continuously free for top-speed calculation. Stretch was also known for its so-called “look-ahead” feature, which enabled it to prepare for a future task even while calculating a present task. The notion of having a computer do many things simultaneously instead of one at a time would become a fundamental design principle of supercomputers in the decades to come.

Stretching the limits of high-performance computing

The Stretch story begins with IBM engineer Stephen W. Dunwell, who had come to notoriety within IBM during World War II, when he helped develop a top-secret computer that could decode intercepted enemy radio messages. In 1954, he coauthored a series of memos about a conjectural computer system dubbed “Datatron” that would “assure IBM a pre-eminent position in the field of electronic data processing” by “making substantial advances on all fronts.” The proposed high-performance computer would take advantage of the new transistor technology.

At the time, IBM was falling behind in scientific computers. A proposal to vault the company to the forefront of high-performance machines caught the attention of both Gene Amdahl, who would later oversee the development of the IBM System/360 computer, and John Backus, who went on to create Fortran, the first widely used, high-level programming language. Project leaders first approached Livermore Radiation Lab in hopes of securing government funding. Ultimately, the IBM team couldn’t create a working model quickly enough, but that initial push opened the door to a subsequent contract with Los Alamos Scientific Laboratory for what became known as Project Stretch, headed by Dunwell.

Dunwell’s team set an ambitious goal to build a computer that was 100 times faster than the IBM 704, the company’s top-of-the-line scientific computer at the time. In the process, Project Stretch initiated research and development efforts into transistors, circuits, memories and power supplies, some of which were later spun off as separate R&D units.

After more than four years, Stretch debuted with the ability to perform more than 30 billion multiplications during a 24-hour period, setting a new standard for high-powered machines that would come to be known as supercomputers. Its use of solid-state transistors made Stretch inherently more reliable than computers that used vacuum tubes, which took up a lot of room and were prone to burning out. But what made Stretch a true breakthrough was its tremendous processing speed, thanks to its ability to calculate different problems in parallel. Stretch could run as many as nine programs simultaneously.

Stretch debuted with the ability to perform more than 30 billion multiplications during a 24-hour period
The Stretch legacy
System/360

Stretch reigned as the world’s fastest computer until being dethroned by a Control Data Corporation supercomputer in 1964. Yet it fell far short of its ambitious commercial goals and was discontinued within a year of its introduction.

IBM had promised to deliver to Los Alamos a computer 100 times faster than the system it was meant to replace — but Stretch topped out at only 30 to 40 times faster. The system’s shortcomings forced IBM to cut the price from USD 13.5 million to USD 7.8 million, which meant the company lost money on every system it sold. Only nine Stretch computers were put into service, and IBM pulled the plug on the system.

Nevertheless, the system would prove to have long-term benefits for IBM and the computer industry at large. Many of the advances Stretch made in computer architecture, circuits, core storage, memory protection and input/output devices were incorporated into the hugely successful IBM System/360, which set IBM on a path to dominate the computer industry for decades. Thirty years after Stretch was introduced, products based on the System/360 architecture accounted for more than half of IBM’s total revenue.

Stretch also led to the development of ever more powerful supercomputers used for everything from weather forecasting and aircraft design to cosmological research and drug development. The processing speed of Stretch was measured in kiloflops, or thousands of floating-point operations per second. The world’s fastest supercomputers are now capable of speeds measured in petaflops, or more than 1 quadrillion (1015) floating-point operations per second.

As IBM chairman and CEO Thomas J. Watson Jr. predicted in 1961, “Stretch is only the beginning in a new decade of supercomputers. Twenty years from now, we will surely look back on it as a relic of the pioneer days.”

Stretch is only the beginning in a new decade of supercomputers. Twenty years from now, we will surely look back on it as a relic of the pioneer days. Thomas J. Watson Jr. IBM chairman and CEO
Related stories The IBM System/360

The USD 5 billion gamble that changed the trajectory of IBM

SAGE

An atomic bomb kicked off the biggest computing project in history, marking IBM’s leadership in a new class of online systems

Breaking the petaflop barrier

IBM Roadrunner smashed an elusive speed record and marked a new era in supercomputing