For two decades the working memory of nearly every computer was a grid of tiny magnetic rings. Each ferrite core is threaded by hair-thin wires and magnetized one way or the other — a single bit that holds its state with the power off. This exhibit is one such plane with nothing hidden: sixty-four cores, their X and Y drive lines, and the sense and inhibit wires woven through them — the arrangement that made ‘core’ a synonym for memory and left us the phrases ‘in core’ and ‘core dump’.
The cleverness is addressing one ring among thousands with just two wires per row and column. Send half the switching current down an X line and half down a Y line, and only the core at their crossing feels enough field to flip; its neighbours, half-selected, are nudged but hold. Here that is real physics, not a lookup — every core runs its own square hysteresis loop, and a flip happens only when the coincident field clears the coercive threshold. Reading is destructive: forcing a core to zero is the only way to ask what it held, and the magnetic collapse of a stored one induces the spike you see cross the sense amplifier. A write-back, vetoed bit by bit by the inhibit wire, then restores what the read destroyed.
Jay Forrester’s coincident-current core, running on MIT’s Whirlwind around 1953, made memory fast, reliable, and cheap enough to build the computer age on; it reigned until the silicon memory chip displaced it in the 1970s. The read-then-restore cycle you can step through here is the ancestor of every RAM access your machine makes — the same arithmetic of fields and thresholds you can wire up gate by gate in DigiSim’s Logic Lab.
Specifications
- Storage
- Ferrite toroid hysteresis — non-volatile, two remanent states
- Addressing
- Coincident X/Y half-currents; only the crossing clears threshold
- Readout
- Destructive sense + inhibited write-back
- Simulated
- Per-core B–H threshold, live sense EMF, real half-select disturb
More from Computing
