Types of Memory
Memory components include the main memory, also known as RAM, and the memory components needed to boot or start a computer, ROM and CMOS. Each of these memory components will be discussed in the following pages. The figure below shows a top-level view of the different types of memory.
Figure 1 RAM, ROM, and CMOS RAM RAM (random access memory) is a temporary holding area for both data and instructions. It is also referred to as main memory. RAM stores data and instructions needed to execute programs. The data in RAM is lost when the computer is turned off. In contrast to accessing data serially, searching sequentially for the data to be retrieved, data in RAM can be accessed directly via its address. Therefore, RAM stands for "random access memory." Random access is similar to accessing a song on a CD directly via its track number, as opposed to finding a song serially on tape. RAM is measured by its memory capacity and latency. Capacity is the maximum number of bits or bytes that can be stored. The capacity of RAM is typically measured in megabytes (MB). Many computers have RAM capacity of 128MB or more. Latency is the delay between the time when the memory device receives an address and the time when the first bit of data is available from the memory device. This delay is also referred to as access time. Latency is typically measured in nanoseconds (ns), billionth of a second (10-9 sec). Latency measures the speed of RAM. Two major categories of RAM called DRAM and SRAM are discussed below. DRAM - Dynamic RAM is a common type of RAM. It is made of an integrated circuit (IC), composed of millions of transistors and capacitors. A capacitor can hold electrons, just as a cup can hold water. An empty capacitor represents a zero, and a non-empty capacitor represents a one. Each capacitor can register either a zero or a one for a memory cell, storing one bit of data. The transistor is like a switch that controls whether the capacitor's state (charged or not charged, 1 or 0) is to be read or changed. Changing the state of a capacitor is like writing new data to a memory cell. However, a capacitor is like a cup that leaks, in order to keep its charge, the memory control needs to be recharged or refreshed periodically. Therefore, it is called the dynamic RAM because its state is not constant. Refreshing capacitors also takes time and slows down memory. There are several types of DRAM, one is SDRAM (Synchronous Dynamic RAM) used in many personal computers. It is fast and relatively inexpensive. It is synchronized to the clock so that data can be sent to the CPU at each tick of the clock, increasing the number of instructions the processor can execute within a given time. Below is a picture of SDRAM configured as a series of DIPs (dual in-line packages), which has two rows of pins that connect the IC circuitry to a circuit board. The entire circuit board shown is called a DIMM (dual in-line memory module).
Figure 2 SDRAM circuit board A faster version of SDRAM is DDR SDRAM (Double Data Rate SDRAM), which transfers twice the amount of data per clock cycle compared to SDRAM. Its capacity is up to 2 GB. Another type of DRAM is RDRAM (Rambus Dynamic RAM), which has a higher bandwidth than SDRAM, but it is more expensive compared to SDRAM. The high bandwidth enhances the performance of applications that access large amounts of data through memory, i.e. real-time video and video editing. You can learn more about RDRAM. SRAM - Static RAM is a type of RAM that uses transistors to store data. Because SRAM does not use capacitors, reading data from SRAM does not require recharging the capacitors. Therefore, it is faster than DRAM. But, because it consists of more electronic parts, it holds fewer bits and costs more compared to DRAM of the same size. SRAM is appropriate for use in the cache because it is fast and cache does not require a large memory capacity. The following chart lists the relative capacity and price of the types of RAM mentioned above.
Table 1 Comparing different types of RAM You can learn more about RAM in the white paper, "A Basic Overview of Commonly Encountered Types of Random Access Memory (RAM)." A table comparing different types of RAM can be found on page 20 of the PDF document. ROM Read-only memory (ROM) is programmed with data hard-wired when it is manufactured. Data and instructions on a ROM are permanent, or nonvolatile, which means they are not lost when power is turned off. Why is ROM needed when RAM enables all the memory operations necessary for a computer? Because data in RAM is lost when the computer is turned off, some instructions are needed for the CPU to start or boot the computer when the computer is first turned on. Therefore, ROM contains a set of instructions that is needed to start the computer. These instructions tell the computer how to access the hard disk, find the operating system, and load it into RAM. Then RAM stores all subsequent instructions to be executed by the CPU. Typically, ROM is hardwired with start-up instructions for the computer. At one time, changing the boot instructions (for example, to integrate a new video card and disable the old one) on a ROM required replacing the entire ROM chip. Currently, a type of ROM, electrically erasable programmable read-only memory (EEPROM), can be updated by applying an electrical field changing instructions stored on the chip one byte at a time. This could slow down the update process. An alternative to EEPROM is flash memory. Flash memory is a type of EEPROM that rewrites data in chunks, usually 512 bytes in size, instead of 1 bit at a time.
|
