CNN acceleration method and system based on OPU

Abstract

The invention discloses a CNN acceleration method and system based on OPU, and relates to the field of CNN acceleration methods based on FPGA. The method includes that an OPU instruction set is defined; the compiler converts the CNN definition files of different target networks, selects an optimal accelerator to configure mapping according to the defined OPU instruction set, and generates instructions of different target networks to complete mapping; the OPU reads the compiled instruction, operates the instruction according to a parallel computing mode defined by the OPU instruction set, and completes acceleration of different target networks. According to the method, the instruction type is defined, the instruction granularity is set, network recombination optimization is carried out, a mapping mode for guaranteeing the maximum throughput is obtained by searching a solution space, and hardware adopts a parallel computing mode; the problem that an existing FPGA acceleration work aims at generating specific independent accelerators for different CNNs is solved, and the effects that the FPGA accelerator is not reconstructed, and acceleration of different network configurations is rapidly achieved through instructions are achieved.

Description

technical field [0001] The invention relates to the field of FPGA-based CNN acceleration methods, in particular to an OPU-based CNN acceleration method and system. Background technique [0002] Deep convolutional neural networks (CNNs) have demonstrated high accuracy in various applications, such as visual object recognition, speech recognition, and object detection, etc. However, its breakthrough in accuracy comes at the cost of high computational cost, which needs to be accelerated by computing clusters, GPUs and FPGAs. Among them, the FPGA accelerator has the advantages of high energy efficiency, good flexibility, and strong computing power, especially in deep CNN applications on edge devices such as speech recognition and visual object recognition on smartphones; it usually involves architecture exploration and optimization , RTL programming, hardware implementation and software-hardware interface development. With the development, people have conducted in-depth researc...

AI Technical Summary

Problems solved by technology

[0004] The object of the present invention is: the present invention provides a kind of CNN acceleration method and system based on OPU, solves existing FPGA acceleration work and aims at generating specific separate accelerator for different CNN, hardware upgrade complexity is high when target network changes, The problem of poor versatility

Images