What Main Memory Is For
A processor can only work on what it can reach quickly. Main memory is the working space where a computer keeps the program it is running and the data that program is using at that moment. It is different from long-term storage, which holds files when the machine is switched off. A common comparison is a desk and a filing cabinet: the desk is small and convenient, the cabinet is larger and slower to get at.
Memory needs two things above all: it must hold on to bits, the on-off values of binary numbers, and any location must be quick to read or write. Many ways of doing this have been tried. The story below follows several major ones in roughly the order they became common, though their use overlapped, and exact dates differ between machines.
Delay Lines
In the 1940s, some early machines stored bits as pulses traveling through a medium. In an acoustic delay line, an electrical signal was converted into a sound pulse that traveled along a tube, often filled with mercury, and was converted back to an electrical signal at the other end. The signal was then fed back in, so the pattern of pulses circulated continuously. EDSAC in Britain and UNIVAC I in the United States are commonly cited as machines that used mercury delay lines.
The idea worked but had drawbacks. Data was available only as it came around, so the machine had to wait for the right bit to pass by. Temperature also affected the speed of sound in the tube, which called for careful control.
Cathode-Ray Tube Memory
Another approach, developed in Manchester in the late 1940s and named for Frederic Williams and Tom Kilburn, used a cathode-ray tube. Writing a dot on the screen left a small electrical charge, and a nearby detector could sense it. Patterns of dots acted as bits. This allowed any location to be read directly, which was an advantage over delay lines. The Williams tube was reportedly used in the Manchester experimental machine and in some commercial computers of the early 1950s. The charges faded and had to be refreshed constantly, and tubes could be temperamental.
Magnetic Drums
A magnetic drum was a spinning cylinder coated with magnetic material, with read-write heads positioned along it. Data was stored as magnetized spots. Drums held more than the other early options, though access still depended on waiting for the right spot to rotate under a head. Because of this they were often used as a middle layer between fast memory and slower storage, and in some machines as main memory itself.
Magnetic Core Memory
During the 1950s, magnetic core memory became the dominant form for general-purpose computers, and it remained widely used into the 1970s. It consisted of tiny rings of magnetic material, each threaded by fine wires. A ring could be magnetized clockwise or counterclockwise to represent a 0 or 1, and electric currents through the wires could set or sense the state. Any ring could be reached directly, and the magnetization stayed in place without power, so core memory retained its contents when the machine was turned off.
Core memory was reliable and fast for its time, but it was also built largely by hand, with wires threaded through rings, which made it costly. Reading a core generally erased it, so the machine had to write the value back immediately.
Semiconductor Memory
The arrival of integrated circuits in the late 1950s and 1960s made it possible to place memory cells on silicon. By the early 1970s, semiconductor memory chips were available, and over the following years they replaced core memory in most new machines. Two main kinds became common:
- Static memory (SRAM): Each bit is held by a small circuit of transistors that keeps its state as long as power is supplied. It is generally fast.
- Dynamic memory (DRAM): Each bit is held as a tiny electrical charge that leaks away, so the memory is refreshed many times per second. It is simpler per bit, which allows more bits on a chip.
Both are volatile: they lose their contents when power is removed. Chip manufacturing methods improved steadily, so memory generally became cheaper and denser, and this trend helped make personal computers practical.
Memory Versus Storage
Several of these technologies blur the line between memory and storage. Drums and tapes were storage media in some machines and memory in others. Today the distinction is cleaner: main memory is fast, volatile and directly accessible by the processor, while storage such as hard disk drives and flash memory keeps data permanently but is accessed more slowly. Many systems also use small amounts of especially fast memory close to the processor, known as cache, to reduce waiting.
Why Memory Design Matters
How much memory a machine had often determined what it could do. Early programmers worked within tight limits and wrote compact code. Design choices about how to organize data, and even the appearance of early video games with their blocky graphics, were shaped by memory limits. Operating systems later added techniques to make limited memory appear larger by shifting data to and from storage as needed.
A Pattern in the History
Looking across these technologies, a pattern appears. Each form solved the problems of the previous one but brought its own, and cost, speed and reliability all pushed the choice. The details changed, but the job stayed the same: give the processor a place to keep bits where it can find them quickly.