Kingston: The Glossary of RAM

The Glossary of RAM

Memory is a computer system's primary workspace. It works in tandem with the CPU, or microprocessor, to store data, programs, and processed information that can be made immediately and directly accessible to the CPU or to other system devices. Memory is central to a computer's operation because it forms the critical link between software and the CPU. Computer memory also determines the size and number of programs that can be run simultaneously, and helps to optimise the capabilities of increasingly powerful microprocessors.

There are many different kinds of memory, each with its own features and benefits. Unfortunately, with so many different types of memory, it can be easy to get them confused. Use this glossary to help sort out the confusion, or as a quick reference to refresh your own memory from time to time.

RAM (Random Access Memory)

Internally, computer memory is arranged as a matrix of "memory cells" laid out in rows and columns, like squares on a checkerboard. Each memory cell is used to store a bit of data, which can be instantaneously retrieved by indicating the row and column location (or address) of the data. Because these bits of data can be individually accessed, retrieved, and modified at random, the type of main memory used in computers is called random access memory (RAM).

RAM is a volatile form of memory, which means that it must have power in order to retain data. When the power is turned off, data in RAM is lost. Contrast this to other storage media such as disks, tapes, and CDs that retain data even without power.

The two main forms of RAM are DRAM and SRAM.

DRAM (Dynamic RAM)

DRAM is the most common type of computer memory. A bank of memory modules using DRAM chips usually forms the core of a computer's main memory. The system uses this memory to temporarily store programs, data, and processed information that moves to and from the processor, video card, and other peripherals.

It is called "dynamic" RAM because it must be refreshed, or re-energised, hundreds of times each second in order to retain data in its memory cells. It has to be refreshed because its memory cells are designed around tiny capacitors that store electrical charges. These capacitors work like very tiny batteries and will gradually lose their stored charges if they are not re-energised.

SRAM (Static RAM)

Static RAM also uses memory cells laid out in rows and columns to store data, but SRAM is about five times faster, twice as expensive, and twice as big as DRAM. SRAM is also volatile (must have power to retain data) but SRAM does not need to be constantly refreshed like DRAM.

SRAM designs use pretzel-like flip-flop circuits that allow electricity to flow through one side or the other depending on which one of two transistors is activated. This "flow through" design is faster than the "stored charge" design of DRAM, but it consumes more power.

Because of its lower cost and smaller size, DRAM is preferred for use in computer main memory, while SRAM, because of its speed, is used primarily for cache memory.

Cache RAM

Cache is a small block of high-speed memory (usually SRAM) located between the CPU and main memory that is used to store frequently requested data and instructions. When the processor needs data, it will check in high-speed cache first to see if the data is there. If not, then the processor will retrieve the data from slower main memory.

Cache works much like a home refrigerator. A refrigerator can be considered a "cache" for groceries. Instead of going to the grocery store (main memory) every time you're hungry, you can go to the refrigerator (cache) first to see if the food you want is there. If it is, then you've saved a lot of time. If not, then you have to spend the extra time to get it from the store.

FPM (Fast Page Mode) DRAM

Most computers in use today use FPM DRAM. This kind of memory is an improvement over older forms of DRAM, making it faster to access data in the same row, or "page". If the data needed is in the same row as the previous data, the memory controller does not have to repeat the row location; it only needs to indicate the next column location. This makes the memory process a little faster.

Using FPM memory is like reading a dictionary. As long as the word you want is on the same "page", it will be easy to scroll down the list and find the definition; but when you have to flip pages, it takes a little longer to find what you want.

EDO (Extended Data Out) DRAM

EDO DRAM is almost the same as FPM, with a slight modification that allows back-to- back memory accesses to occur much faster. Because EDO is easy to implement, it has gained wide acceptance in a very short span of time.

A computer system must be designed to support EDO in order to make use of the extra efficiency it offers. EDO memory in a system that doesn't support it will work, but there will be no performance increase. At present, FPM and EDO DRAM make up the majority of main memory for computers.

EDO is sometimes called Hyper Page Mode DRAM.

BEDO (Burst EDO) DRAM

Burst EDO is an innovation on standard EDO that allows a "burst", or series, of data to be transmitted from memory with a single request. The assumption behind this feature is that the next data-address requested by the CPU will be sequential to the last one, which is usually true.

In BEDO DRAM all memory accesses occur in bursts.

SDRAM (Synchronous DRAM)

SDRAM is another form of memory developed shortly after EDO. This technology is a more radical innovation that can synchronise itself with the system clock that controls the CPU. Being "in sync" with the processor eliminates timing delays and makes the memory retrieval process much more efficient.

In order to use SDRAM, a computer system must be designed to support that kind of memory. Most computers shipping today support EDO, but a few are beginning to support SDRAM. Over the next few years, FPM will phase out, and SDRAM will most likely overtake EDO as the standard memory for high-performance systems.

SGRAM (Synchronous Graphics RAM)

SGRAM is an extension of SDRAM that includes graphics-specific read/write features. SGRAM allows data to be retrieved and modified in blocks, instead of individually. This reduces the number of reads and writes that the memory must perform, and increases the performance of the graphics controller by making the process more efficient.

RDRAM (Rambus™ DRAM)

RDRAM is a completely unique design developed by Rambus, Inc. RDRAM is extremely fast, but requires significant changes in the memory controller and memory/system interface to use. RDRAM uses a narrow, high-bandwidth "channel" to transmit data at speeds about ten times faster than standard DRAM. At present, RDRAM is being used in some game machines and PC graphics applications.

VRAM (Video RAM)

Graphics memory must work very quickly to update, or refresh, the screen (60-70 times a second) in order to prevent screen "flicker". At the same time, graphics memory must respond very quickly to the CPU or graphics controller in order to change the image on screen. With ordinary DRAM, the CRT and CPU must compete for a single data port, causing a bottleneck of data traffic.

VRAM is a "dual-ported" memory that solves this problem by using two separate data ports. One port is dedicated to the CRT, for refreshing and updating the image on the screen. The second port is dedicated for use by the CPU or graphics controller, for changing the image data stored in memory.

VRAM works much like a fast food drive-through that uses two windows. After you place an order, you pay at one window, then drive up and get your food at the next window. This makes the process faster and more efficient.

WRAM (Window RAM)

WRAM is another type of dual-ported memory also used in graphics-intensive systems. It is slightly different from VRAM in that its dedicated display port is smaller, and that it supports EDO features.