Graphics processing unit (GPU) program optimization method based on compute unified device architecture (CUDA) parallel environment

Abstract

The invention relates to a graphics processing unit (GPU) program optimization method based on compute unified device architecture (CUDA) parallel environment. The GPU program optimization method defines performance bottleneck of a GPU program core and comprises global storage access delay, shared storage access conflict, instruction pipelining conflict and instruction bottleneck according to grades. An actual operational judgment criterion and a bottleneck optimization solving method of each performance bottleneck are provided. A global storage access delay optimization method includes transferring a shared storage, access merging, improving thread level parallelism and improving instruction level parallelism. A shared storage access conflict and instruction pipelining conflict optimization method includes solving bank conflict, transferring a register, improving thread level parallelism, and improving instruction level parallelism. The instruction bottle neck includes instruction replacing and branch reducing. The GPU program optimization method provides a basis for CUDA programming and optimization, helps a programmer conveniently find the performance bottleneck in a CUDA program, conducts high-efficiency and targeted optimization for the performance bottleneck, and enables the CUDA program to develop computing ability of GPU equipment to the great extent.

Description

technical field [0001] The invention relates to a parallel computing and data processing method in the fields of graphics, animation, scientific computing, geology, biology, physical simulation, etc., in particular to a GPU kernel program optimization method based on CUDA architecture, which belongs to the field of high-performance computing. Background technique [0002] The CUDA architecture (Compute Unified Device Architecture) is a parallel computing architecture for GPU (Graphic Processing Unit) graphics processors and other devices. CUDA C, C++, OpenCL, RapidMind, etc. CUDA C is a C language extension based on the CUDA architecture. Programmers can easily use this set of APIs for GPU programming. The realization of the effect of the program depends on the programmer writing a CUDA kernel program with high performance, stable function and strong portability. The CUDA kernel program is also called the kernel function, which is a parallel computing function running on th...

AI Technical Summary

Problems solved by technology

[0007] The above-mentioned public optimization technologies are limited to several aspects of CUDA program optimization, which are not enough to fully utilize the computing power of the GPU device, and the actual optimization effect is not perfect; at the same time, these documents or authors often do not give the technical theory Elaboration, unclear explanation of the use occasions and conditions of optimization technology, so it is difficult to achieve practical and operable standards

Images