Generated by DeepSeek V3.2| magnetic core memory | |
|---|---|
| Name | Magnetic Core Memory |
| Caption | A module of magnetic core memory, showing the grid of wires and tiny ferrite rings. |
| Inventor | An Wang, Jay Forrester |
| Derived from | Williams tube, delay-line memory |
| Succeeded by | Semiconductor memory |
| Date | 1949–1950s |
magnetic core memory was the predominant form of random-access memory for computers from the mid-1950s until the mid-1970s. It stored data as the magnetic polarity of tiny ferrite rings, or cores, which were woven into a dense grid of wires. This technology was non-volatile, reliable, and became the standard for mainframe and minicomputer memory, playing a critical role in the advancement of computing during the Space Race and the Cold War.
Early computer memory systems like the Williams tube and delay-line memory were volatile and unreliable. The foundational concept for a more stable magnetic memory was independently developed by several researchers, including An Wang at Harvard University and Jan A. Rajchman at RCA. However, the practical, coincident-current system that made large-scale production viable was invented by Jay Forrester at the Massachusetts Institute of Technology as part of the Whirlwind I computer project, which was funded by the United States Navy. This development was crucial for the Semi-Automatic Ground Environment air defense system. Key manufacturing patents were held by IBM, which licensed the technology widely, while companies like Digital Equipment Corporation and Data General used it extensively in their popular PDP-8 and Nova minicomputers.
Each small ferrite core acts as a toroid and can be magnetized in one of two directions, representing a binary '0' or '1'. The cores are arranged in a planar grid, with wires—typically four—threaded through each core. To write a bit, electrical currents are sent through a specific X line and Y line; the combined magnetic field at the intersection switches the polarity of that single core. Reading is destructive: a current pulse attempts to set the core to a '0' state; if the core was already a '0', a small voltage is induced on a third, sense line, but if it was a '1', the flipping generates a larger, detectable signal. A fourth inhibit line is used during write operations to prevent unwanted core switching.
The cores themselves were made from ferrite, a ceramic magnetic material with a rectangular hysteresis loop, which provided distinct and stable magnetic states. They were manufactured by companies like Ferroxcube and were extremely small, often just millimeters in diameter, requiring precise, often manual, assembly. Workers, frequently women in facilities like those at IBM, would weave the fine copper wires through the cores using specialized tools and microscopes to create planes. These planes were then stacked to form a three-dimensional array, with each plane representing one bit of a word, and the stack was potted into a solid module for physical protection.
This memory technology offered a compelling balance of speed, reliability, and cost for its era. Typical cycle times ranged from a few microseconds down to under one microsecond in later, optimized designs. Its key advantage was non-volatility; data remained intact even if power was removed, a critical feature for many early systems. It was also highly reliable compared to its predecessors, being immune to environmental disturbances and possessing a long operational life. However, it was relatively expensive to manufacture, consumed more power than later technologies, and was physically bulky, limiting its ultimate storage density.
Magnetic core memory was the central memory technology in nearly every significant computer from the late 1950s through the 1970s, including the IBM System/360, the Apollo Guidance Computer, and machines from UNIVAC and Control Data Corporation. Its reliability was essential for real-time systems in aerospace, defense, and industrial control. The technology was ultimately displaced by the advent of cheaper, faster, and denser semiconductor memory, specifically dynamic random-access memory integrated circuits, in the early 1970s. Today, it is remembered as a foundational technology of the computing revolution, with modules preserved in museums like the Computer History Museum and the National Museum of American History. Category:Computer memory Category:Early computers Category:Obsolete computer hardware