基于GPU的矩阵求逆性能测试和分析  

Performance Testing and Analysis for Matrix Inversion Base on GPU

作　　者：刘丽[1] 沈杰[2] 李洪林[2] 

机构地区：[1]华东理工大学信息科学与工程学院,上海200237 [2]华东理工大学药学院,上海200237

出　　处：《华东理工大学学报（自然科学版）》2010年第6期812-817,共6页Journal of East China University of Science and Technology

基　　金：国家"973"计划基金项目(2009CB918501);国家自然科学基金项目(20803022)

摘　　要：在CPU串行运算模式下实现大规模矩阵求逆是一个非常耗时的过程。为了解决这一问题,基于NVIDIA公司专为GPU(图形处理器)提供的CUDA(计算统一设备架构),从新的编程角度出发,利用GPU多线程并行处理技术,将矩阵求逆过程中大量的数据实现并行运算,从而获得了较大的加速比。同时,根据程序的执行结果,分析了GPU的单精度与双精度的浮点运算能力及其优、劣势。最后,通过分析数据传输时间对GPU性能的影响,总结出适合GPU的算法特征。For the CPU serial operation mode,it is a very time-consuming process to obtain the inverse of large-scale matrix.Aiming at the above shortcoming,this paper proposes a new programming method based on the common platform CUDA for GPU designed by NVIDIA.By using the multi-threaded parallel processing technology of GPU,a large scale of data during solving the inverse matrix are parallelly computed such that a higher speedup may be obtained.Moreover,both the single-precision and the double-precision FLOPS of GPU are analyzed according to the results of this program.Finally,some characteristics of the proposed algorithms are summarized by analyzing the effect of the data transmission time on the performance of GPU.

关 键 词：图形处理器(GPU) 计算统一设备架构(CUDA) CPU 并行运算 矩阵求逆 

分 类 号：TP311.11[自动化与计算机技术—计算机软件与理论]