The IBM 7090: The Supercomputer That Launched the Space Age
Introduction
In 1959, IBM released the 7090 — a transistorized computer that was fast enough to compute orbital trajectories, powerful enough to launch satellites, and influential enough to create the Unix operating system. It was one of the most important machines of the 20th century, and its legacy extends far beyond its own lifetime.
The IBM 7090 was the successor to the IBM 709 — a smaller, less powerful machine that was also transistorized but not widely adopted. The 7090 was designed from the ground up as a high-performance scientific computer, with a 36-bit word, magnetic core memory, and a design that balanced speed with reliability.
By 1961, NASA had ordered several 7090s for computing Apollo trajectories. By 1962, the first artificial satellite launched by the U.S. was guided by a 7090. By 1969, the Apollo 11 lunar landing calculations were being processed by a 7090 at NASA's Goddard Space Flight Center.
The 709 vs. the 7090: Evolution of a Supercomputer
The IBM 709 was IBM's first transistorized computer, released in 1958. It was a significant improvement over the vacuum-tube IBM 704 it replaced — faster, more reliable, and more compact. But the 709 had limited memory (only 4K words of core memory) and was primarily designed for scientific calculations.
The IBM 7090 was designed as the 709's successor — a more powerful, more flexible machine with 32K words of core memory (eight times more than the 709), faster processing, and improved I/O capabilities. It used the same 36-bit word length as the 704 and 709, ensuring compatibility with existing software.
The 7094 was a faster version of the 7090, released in 1961, with improved instruction timing and additional features. The 7094 became the preferred model for most scientific and government installations.
Architecture: 36-Bit Word and Magnetic Core Memory
The IBM 7090 used a 36-bit word length — each word could hold one floating-point number or four 9-bit characters. This was a deliberate design choice: 36 bits was long enough for precise scientific calculations but short enough to be processed efficiently by the transistorized circuits of the time.
The memory was magnetic core — small rings of magnetic material arranged in a grid, each ring representing one bit of memory. The 7090 had 32K words of core memory (about 128KB of storage), which was enormous for 1959. Each word could be accessed in about 1 microsecond — extremely fast for the era.
The instruction set included basic arithmetic (addition, subtraction, multiplication, division), logical operations, and floating-point operations. The 7090 could perform about 10,000 additions per second — fast for 1959, but slow by today's standards.
The Space Program Connection
NASA was one of the largest customers of the IBM 7090. The 7090 was used to compute orbital trajectories for the Mercury, Gemini, and Apollo programs. The most famous program was PAI (Position and Attitude Integrator), which computed the trajectory of the Saturn V rocket from launch to lunar orbit.
Each launch required hundreds of trajectory calculations, each taking about 45 minutes on the 7090. The calculations were done on punch cards: the input program and data were loaded from punch cards, the output was written to punch cards, and a separate printer converted the cards to readable text.
The 7090's reliability was critical for the space program. Unlike modern computers that could be restarted in seconds, the 7090 was a large machine that took minutes to warm up and stabilize. When it worked, it worked reliably. When it failed, the entire calculation had to be restarted.
The Bell Labs Connection: Unix Was Born on a 7090
The 7090's most unexpected legacy was in software. In the late 1960s, a team at Bell Labs — Ken Thompson, Dennis Ritchie, and others — began developing an operating system on a GE-635 computer. When the GE-635 was retired, they moved the project to an IBM 7090 at Bell Labs.
On the 7090, they wrote the first version of Unix — a small, efficient operating system designed to be used by programmers. The 7090's 36-bit architecture influenced Unix's word size, and the machine's reliability allowed the team to develop and test their operating system without interruption.
The Unix that was created on the 7090 was the ancestor of every modern operating system: Linux, macOS, iOS, Android, and even Windows have Unix roots. The 7090, a supercomputer from 1959, is the machine where Unix was born.
Performance and Legacy
The IBM 7090 was one of the first computers to be called a "supercomputer" — a machine that was significantly more powerful than the computers of its time. It was also one of the first machines to be used for a wide range of applications: scientific computing, business data processing, space exploration, and software development.
IBM built about 89 7090/7094 machines between 1959 and 1963. Each one was a significant investment — each machine cost between $4 million and $5 million (about $40-50 million today). But each one was also a workhorse — used for everything from weather forecasting to stock market analysis to computing the first artificial satellite trajectory.
The 7090's architecture — particularly its 36-bit word length and magnetic core memory — influenced the IBM 360 line and the broader industry. The 360's design philosophy — a family of compatible machines with a common instruction set — was a direct result of the experience with the 7090/7094 family.
Key Takeaways
The IBM 7090 was the bridge from the punched-card era to modern computing. It was one of the first supercomputers, one of the first machines to be used for space exploration, and one of the first machines where Unix was born. Its 36-bit word, magnetic core memory, and reliable transistorized architecture made it the template for the IBM 360 line and the broader industry.