Random-access memory, acronym RAM, English for memory with random access, is a computer memory, each memory location of which can be accessed at the same speed (in contrast to, for example, a hard disk that has to wait until the relevant spot under the read head spins). With RAM, data can also be read from the memory in an arbitrary (random) order or data can be written into the memory. The term “RAM” is also used, which is however a pleonasm since the “M” already stands for memory.
Read-only memory (ROM) is a read-only RAM: the data in a ROM cannot be changed.
After the drum memory came working memory in the 1960s in the form of magnetizable rings (ring core memory). The English term core (= core), which means working memory, harks back to this. This was RAM. It consisted of a number of rings of magnetizable material that were strung together with three or four wires to form a rectangular mesh. Informally, it was therefore referred to as a mat. By reading the memory, the content was lost (destructive reading), so it had to be written back again. When the computer was turned off, the contents of the memory were retained (non-volatile memory).
Transistors
Transistor circuits were used at the same time for faster storage. Usually two transistors (a so-called flip-flop) were on a single printed circuit board, and a single bit could be stored on them. This precious memory was used for the internal registers of the CPU. Content was lost when the computer was turned off (volatile memory).
Memory chips
During the 1970s, integrated memory circuits entered the market. An integrated circuit contains transistors and is thus no different in operation from the memory type discussed above. However, it is more compact, more robust and cheaper. In 1974 it was thought that it would never be possible to make more than 8 kiB on one chip (“the chip burns itself to death in minutes”), in 2003 there are chips on the market that can each contain a GiB. (1 GiB = 1024 MiB = 1048 576 kiB = 1 073 741 824 bytes = 8 589 934 592 bits.) Smaller 8GiB chips are now being “stacked” on one large chip. They are then, as it were, glued together. This means that there can be 32 or even 64 chips in one housing.
Dynamic and static memory chips
The working memory of the modern computer consists of two types of RAM: DRAM (dynamic RAM), and the even faster but more expensive SRAM (static RAM).
A bit in dynamic RAM effectively consists of a small capacitor that must be refilled a number of times per second in order not to lose its contents. Only one transistor is needed for each memory cell to store that one zero or one.
Each bit in static RAM consists of a transistor circuit that independently remembers its state as long as there is a supply voltage. The transistor circuit that can memorize a zero or one is called a flip-flop, and it requires at least two transistors.
SRAM is mainly made in CMOS technology today. Here 6 transistors are used per memory element. An inverter is made with two transistors. Two inverters together form the actual memory element. Finally, the last two transistors are used to address the bit.