Central Processing Unit (Cpu)
The central processing unit (CPU) is the component of a computer that actually processes the data. CPU designs vary a great deal, but for our purposes we may treat all CPUs as having simply an arithmetic-logic unit, a control unit, and several registers. In recent designs, a memory management unit is also often found.
In early days the various parts of a CPU were located in separate physical devices, so the computer as a whole had to be quite bulky. Since Intel came out with the 4004 microprocessor in the 1970s, however, the entire CPU has been located on a single silicon chip. Any chip that holds a complete CPU is called a microprocessor; indeed, today the words "CPU" and "microprocessor" are virtually synonymous. The complexity of chips has increased tremendously in the last 30 years. For instance, Intel's 4004 contained only about 29,000 transistors, while the Pentium 4, released in 2001, contains on the order of 42 million transistors.
The arithmetic-logic unit (ALU) of a CPU, as its name implies, performs arithmetic and logical operations. Arithmetic operations involve the basic elements one learns in grade school: addition, subtraction, multiplication, and division. However, the ALU does not work with regular decimal (base 10) numbers, but processes numbers in binary (base 2) form. Logical operations (in this context) involve comparing numbers to see which is larger or smaller.
Each CPU has a built-in clock that controls the pace of events and synchronizes the operations of its various parts. This clock is a specialized electronic circuit that generates a very rapid electrical pulse, typically at a rate of hundreds or thousands of megahertz (MHz, millions of cycles per second). The rate at which this pulse is generated on a CPU is called its clock speed, while a single pulse is called a clock tick. The clock on the CPU is not the same as the computer's system clock, a device that keeps real time and date for the benefit of human users; in almost all computers, there is a separate circuit for the system clock.
The control unit of a CPU coordinates all the actions performed by the CPU and the peripheral devices it controls. The control unit performs four basic functions: fetch (getting the next program instruction from memory), decode (determining what the instruction just fetched means), execute (performing the requested action), and write-back (writing the results, if necessary, to memory or to a local register). This whole four-event cycle is called a machine cycle or processing cycle and is said to consist of two parts: the instruction cycle (fetch and decode), and the execution cycle (execute and write-back). A modern CPU can go through many millions of machine cycles per second. In general, a machine cycle takes more than one clock tick to complete, but a technique called pipelining, whereby the next cycle is begun while one (or several) is ongoing, increases speed. Pipelining continues to be a major area of study in CPU design.
A CPU is capable of executing a set of basic operations, such as writing a character to memory, reading a character from memory, or comparing two numbers to see which is larger. The code word ordering a CPU to perform one of these basic operations is called an instruction, and the set of instructions that the CPU is able to perform is called its instruction set. Different microprocessor chips have different instruction sets. Software applications designed to run on a computer with a certain type of CPU must be created to use its instruction set, and software cannot be used on a CPU that has a different instruction set than the one the software is written for. Thus, for example, a software program written for the Macintosh PowerPC chip will not run on a Dell computer based on an Intel chip.
There are two major types of CPUs, which are classed according to the instruction-set philosophy adopted by their designers. The first type, called a CISC (complex instruction set computer), provides programmers with many instructions, and with CPU hardware that supports many complex operations directly without needing them to be broken down in software into multiple steps. CISC chips such as the Motorola 68040 or the Intel Pentium make the programmer's life much easier, but are themselves more physically complex and thus more difficult to design and verify. CISC chips often consume more power, which may be a factor in computers where power is supplied through a battery or where heat dissipation is a concern, and they may give lower performance on average than other chips having the same clock rate. (The Pentium series chips are only half as fast, or worse, compared with PowerPC chips of comparable clock speeds.)
The other type of processor is the reduced instruction set (RISC) computer. RISC machines supply sparser instruction sets, making such chips easier to design and verify and less expensive to manufacture. Such chips also consume less electrical power. However, programmers need to work harder when using these chips to perform complex tasks. Examples of RISC chips include the PowerPC, the Silicon Graphics Inc. MIPS, and the Hewlett-Packard PA.
A memory management unit (MMU) is present in most large modern computer systems, sometimes as a component of the CPU and sometimes as a separate component. The MMU may be thought of as a lens through which programs running on the CPU observe the memory. As with any lens, we can talk about the real world and the image of the world as seen by an observer looking through the lens, that is, the virtual image. This metaphor is the basis of the term "virtual memory." There are quite complex MMUs in CISC architectures such as the Motorola 68040 and the Intel Pentium, in keeping with the CISC philosophy of hardware extravagance. The memory management units on modern RISC architectures are generally much simpler, and usually heavily rely on system software to perform many of the more complex MMU functions.
This is the complete article, containing 988 words
(approx. 3 pages at 300 words per page).
