Sharing of computer-related components ~

The components of a computer can be divided into three parts: the central processing Unit (CPU), the main memory and the input/output system.

In fact, it can also be added to the Part IV, is the system bus, the system bus as a connection to each device, its design is not so simple.

In this paper, we first briefly introduce the components, the opposite of some sub-functional modules specific deployment, according to the circumstances of the separate written.

The CPU is divided into three parts: the Arithmetic logic operation unit (ALU, Arithmetic logic unit), the control unit, the register.

The ALU here is the main body of the data execution logic, shift, arithmetic operations.

Registers are high-speed, independent storage units used to hold temporary data, and the registers in the CPU are the fastest-reading memory in the computer and, of course, the most expensive. So can not be used in large-scale, only in the CPU as a register, not as main memory.

1) data Register (R1~RN): As the name implies is the storage of data, including input data and operation results; In the past, only a few registers were used to store the input data and the results of the operation, and now, as more and more complex operations are implemented by hardware devices (rather than using software), So the computer uses dozens of registers in the CPU to improve the operation speed. And some registers are required to hold the intermediate results of these operations.

2) instruction register (ir,instrution register): The CPU takes out the instruction from memory and stores the instruction in the instruction register, explaining and executing the instruction.

3) Program counter (Pc,programm counter): The program counter holds the currently executing instruction. When the current instruction is executed, the counter will automatically add 1 to the memory address of the next instruction.

The control unit controls the operation of each subsystem and is controlled by signals from the control unit to other subsystems.

The main memory is the second subsystem in the computer, it is a collection of storage units, each storage unit has a unique identity, called the address. The data is made up in memory with bits called words, which can be simply understood as the underlying transmission unit in memory. From memory to the CPU read data, then is a bits, or two two, or 16 16, this is to be based on the original design of the designer scale, the use of scale to consider, but anyway, will eventually form a memory common "word" size scale, For example, some are using 8-bit binary as a word, then each transmission in the bus is a 8-bit binary digital, the word is a basic unit. Often the memory unit is also designed, for example, a word has 8 bits, because each storage unit is formed by the 8 trigger gate circuit.

It is necessary to access each word in memory with a corresponding identifier. Although programmers use naming to differentiate between words (or a combination of words), each word is identified by an address at the hardware level. The total number of independent address units identified in memory is called the address space.

Memory addresses are defined with unsigned binary integers.

Input/output system:

The input/output subsystem in the computer represents a series of devices that enable the computer to communicate with the outside world and store programs and data in the event of a power outage. Input/output devices can be divided into two categories: non-storage devices and storage devices.

1) Non-storage devices, is those mouse ah, keyboard ah, monitor Ah, printer ah, and so on.

2) memory Device: For example, disk, tape, CD and so on. This class can be separated from the memory in a separate article in terms of.

Bus:

The CPU and memory are usually connected by a three-ancestor line called bus, respectively: Data bus, address bus, control bus.

The input/output device is not directly connected to the bus and requires a layer of intermediate equipment, namely the controller, or interface. Because the input/output devices are not necessarily electronic devices, are some electromechanical, magnetic, optical equipment, etc., its processing speed is much slower, and CPU, memory and other different steps.

There are many types of interfaces: SCSI, FireWire, USB,HDMI, for example.

Normally the CPU uses the same bus to read and write data between main memory and input/output devices, the only difference being the instruction. There are two ways: I/O stand-alone addressing and I/O memory mapping addressing.

In I/O stand-alone addressing, the instructions used to read and write memory are completely different from the instructions used to read and write the I/O device, and depending on the instruction, the CPU naturally knows exactly where to read the data, and in the I/O memory mapping addressing, the CPU will enter/ Each register in the output controller is considered to be a storage word in memory, that is, to the CPU, it can not or do not care where it is to read the data, it appears to be in the same place (main memory) to read data, this way has the advantage that there is a small instruction set, because not for i/ o The device creates the instructions separately.

A general-purpose computer uses a series of instructions called programs to process data. The CPU uses repetitive machine cycles to execute instructions in the program, one step at a time. Includes three steps: Take instruction, decode and execute.

Because the I/O device has a much slower running device than the CPU, the operation of the CPU must in some way be synchronized with the I/O device, and three ways are designed for synchronization, namely:

1) Program Control I/O

The CPU periodically queries the status of I/O devices

2) Interrupt control I/O

When the I/O device is ready, it notifies (interrupts) the CPU.

3) Direct Memory access (DMA)

This method requires a DMA controller to assume some of the CPU's functions. There are registers in the DMA that can save the data block before and after the memory transfer. When the data is ready to be transferred, the DMA controller notifies the CPU that it needs to obtain the bus's right to use.

Different architectures:

1) CISC, complex instruction set computer (complex instruction set computer), the CISC architecture design strategy is to use a large number of instructions, including complex instructions.

Intel has developed a Pentium series processor that uses the CISC architecture.

2) RISC, simplified instruction set computer (reduce instruction set computer), complex instruction using a simple instruction subset simulation.

3) assembly line

This is the way to change the throughput (the total number of instructions completed per unit of time), which is the idea that if the control unit can perform two or three stages simultaneously, the next instruction can begin before the next instruction is completed.

4) Parallel processing

The same is the improved throughput.

SISD Organization (single instruction flow, single stream), SIMD organization, MISD organization, MIMD organization (multi-instruction flow, majority data stream).

As above is simply a description of the composition of the computer, where each block is worth opening alone, such as memory Ah, bus structure ah, I/O and CPU before the connection, I/O interface, serial processing/parallel processing mode.

Sharing of computer-related components ~