Introduction
The processor includes a central processing unit, a main memory, an input-output interface, and the addition of peripheral equipment constitutes a complete computer system. A processor is a component that processes programs and data stored in a computer system and executes instructions in accordance with the steps specified by the program. Algorithm processor refers to a processor used to run complex algorithm programs or a processor dedicated to running various algorithms or special algorithm programs.
Neural processor
A non-traditional computer based on the principle of artificial neural network, also known as neural processor. The composition principle and working method of a neural computer are completely different from traditional computers or symbol processors. The symbol processor simulates the information processing function of human abstract thinking based on the symbolic concept, while the neural processor simulates the information processing function of the adaptive adjustment of the stimulus response at the level of the biological neural network microstructure. A neural computer composed of a neural network is generally composed of many processors or neural-like nodes. Each node is connected to many nodes, and each node has an activation energy level that can be changed. When the activation energy level of a node reaches a certain level, it will transmit a signal to other nodes connected to it. The strength of the signal is the combined result of the activation energy level and the connection strength. Neural computer is a system that simulates the information processing function of the human brain through parallel distributed processing and automatic organization. It has learning-automatic organization, thinking, reasoning, memory, large-scale operations, target and image recognition, signal processing, prediction, classification, etc. Function. Its appearance is due to the following two reasons: On the one hand, it is the development of component technology that plays a decisive role in the structure of electronic computers; on the other hand, it is the demand for faster and more effective solutions to practical problems.
Array Processor
According to the principle of parallel computing, in order to achieve high-speed computing, multiple processors are arranged in an array form. The array processor is an array composed of multiple processing units (PE) under the control of a single control unit to process the same instruction in parallel. According to the Flynn classification, it is a single instruction stream and multiple data stream computer, and the multi-processor structure is composed of several processors (or computers), each of which can execute its own program independently. The processors in the multiprocessor system are interconnected in a certain form to realize data exchange and synchronization between programs. The processing object of the array processor is an array, not a single operand. With the development of computational mathematics, matrix calculation has appeared, and its application range is very wide. In addition to general matrix operations, there are matrix eigenvalue calculation, linear programming, table look-up technology, digital filtering technology, Fourier analysis, etc. The array machine has several processors arranged in a square array and corresponding operation control devices, storage devices and input/output devices. For example, 64 processors are arranged in an 8×8 array connected to each other. When performing matrix operations, several independent processors solve different problems at the same time, or solve different parts of the same problem at the same time. Each processor executes the same instruction at the same time and exchanges data with neighboring processors at a speed of hundreds of millions of times per second.