LIS Course Notes - Magnetic Core Memory

Computing History - Magnetic Core Memory

This page features some pictures of a 10,000 Bit magnetic core memory. After relays and vacuum tubes, but before RAM memory, there was core. Even today, core leaves its legacy. An inelegant program which uses a lot of RAM is occasionally called a 'core hog'. When programs abended (i.e. crashed), the operating system would print out the contents of core for diagnosis. In the UNIX and Linux world, the diagnostic file containing the memory dump is still named "core". That's why someone who talks too much is sometimes nicknamed 'Coredump'. Many thanks to Peter and Susan Wesoly for preserving this bit of computing history.


Magnetic cores. Voltage applied along the grid wires changed the polarity of the cores.	25 rows of 50 cores each times 8 planes = 10,000 magnetic cores.	Each core represents a single Binary Digit (Bit).

Soldered jumpers are partially visible at the top..	The eight planes shown from the side.	Wires were tied in bundles using string.

For an explanation of the underlying physics, see "Magnetic fields and how to make them", by Andrew Duffy of Boston University. Thanks to Keith Thomas for suggesting that the machine for which this particular memory was most likely designed, may have used a 24-bit word length, with 1 parity bit for error detection and correction, hence all the multiples of 25.