Feedforward neural network hardware realization method based on multicore technology

Abstract

The invention discloses a feedforward neural network hardware realization method based on multicore technology, relating to a computer system based on a specific calculation model. The invention adopts an NIOS II soft core processor and utilizes Quartus II software in an FPGA development system thereof to build a multicore processor system, the number of processor cores is only limited by chip scale, each processor core can be independently programmed, and multiple processor cores can simultaneously run respective program and simulate the characteristics of nerve cells of any kind by performing programming of one single core; in the network studying process, the single cores calculate output amount in a parallel mode and finish the adjustment of corresponding weights and thresholds; the input and output of each single core are performed with data exchange via a multiport memory to stimulate feedforward neural network with any structure, which realizes the purposes of high speed, distribution and parallel calculation of the neural network.

Description

technical field [0001] The invention relates to a computer system based on a specific calculation model, in particular to a hardware implementation method of a feedforward neural network based on multi-core technology. Background technique [0002] At present, the application method of artificial neural network is mainly based on the software simulation of general-purpose computer. Although this implementation method has the advantages of flexible application and no need for special hardware, the main problem is: because the general-purpose computer runs the program according to the order of instructions, so , it is impossible to truly simulate the characteristics of high-speed, distributed, and parallel computing of biological neural networks. However, the implementation methods based on pure hardware, such as the implementation methods using analog circuits, digital circuits or hybrid circuits, can only simulate specific networks, and it is difficult to simulate the weight...

AI Technical Summary

Problems solved by technology

[0002] At present, the application method of artificial neural network is mainly based on the software simulation of general-purpose computer. Although this implementation method has the advantages of flexible application and no need for special hardware, the main problem is: because the general-purpose computer runs the program according to the order of instructions, so , cannot truly simulate the characteristics of high-speed, distributed, and parallel computing of biological neural networks

However, the implementation method based on pure hardware, such as the implementation method using analog circuit, digital circuit or mixed circuit, can only simulate a specific network, and it is difficult to simulate the weight storage and structure adaptation of the neural network.

In addition, by combining existing microcontrollers, such as DSP, single-chip microcomputers, etc., the method of simulating neural networks is expensive and the circuit scale is large, and large-scale neural network simulation cannot be realized.

Images