Bits and Bytes
Grouping together 8 bits forms what we call a "byte". A byte is the most basic value of computer information. Virtually every value of measurement in computing is done in bytes. The only other time that bits are used is when measuring network data rates, but again, for ease of communication, we tend to convert those amounts into byte values. When talking about memory it's important to understand and know how to convert bits into bytes. Simply divide the bit number by eight, and you have the byte count.
Computer memory is sold retail by its Megabyte (MB) density, but is rated according to its bit density. Some manufacturers are nice enough to print the MB size right on the memory stick, others simply mark the memory with its bit arrangement (which ends up looking like cryptic math), so it helps to know how to convert bits to bytes.
Semiconductor memory stores information in bit cells, which are arranged by bit pattern, and each bit cell is given a unique address so that the computer can keep track of what information goes where. There are two basic types of memory cells, "Static RAM" and "Dynamic RAM."
Static RAM, or SRAM, uses the electronic equivalent of a light switch to store data. When you turn on a light switch, the switch stays on, and power flows through the switch until you either flip the switch to the OFF position, or turn off the power to the circuit. SRAM memory works in exactly this manner. Once the state of a switch is set, the cell keeps that information until the switch is reset by a special reset signal, or until power to the device is removed.
SRAM memory has the benefit of being very fast, requiring very little support circuitry to operate, thereby making it the memory type of choice for cache applications. The drawback to SRAM memory is that it's very expensive, and cannot be packed as densely as DRAM. Remember those SRAM cache chips on the Katmai P3 processor? Those two chips only added up to 512KB. Compare that to a typical DRAM chip which can store 8MB.
Dynamic RAM, or DRAM uses a capacitor cell to store data. A capacitor is essentially a tiny, temporary battery. Once charged, a capacitor begins to lose its energy along an exponential curve (the rate of energy loss is not linear, or straight). As such, DRAM requires special support circuitry to keep the capacitors charged in order to retain its data. How often the memory is recharged is called the refresh rate. If the charge on the capacitor drops too much, the information stored there might be lost. This requires the memory to be refreshed on a regular basis to prevent the loss of information. When the memory is being refreshed, it cannot be accessed to retrieve information, this is called latency, the lag time in between allowable accesses into memory.
As compared to SRAM, Dynamic RAM can be packed a lot more densely, which means that a DRAM chip can be made to store more data than an SRAM chip. However, since DRAM needs to be constantly recharged, this makes it slower to read and write information to and from the memory. DRAM is suitable for applications where a large amount of storage is needed, such as main system memory.