CUDA并行计算的应用研究  

Study on application of parallel computation on CUDA

作　　者：董荦[1] 葛万成[1] 陈康力[1] 

机构地区：[1]同济大学电子与信息工程学院,上海200092

出　　处：《信息技术》2010年第4期11-15,共5页Information Technology

基　　金：获德国罗德施瓦茨公司资助

摘　　要：统一设备架构(CUDA)是NVIDIA公司提出的一个基于GPU通用计算的开发环境,它针对GPU多处理单元的特性,通过并行计算提高大规模运算的速度。根据CUDA技术的特点,提出了基于CUDA的并行图像锐化、中值滤波和字符搜索算法,并论述其关键技术和基本执行流程。试验结果表明,提出的方法相对于CPU方法在运算速度上有不同程度的提高和下降。这同时体现了CUDA的优势和局限性,为其更复杂的应用提供了参考和依据。Compute Unified Device Architecture （CUDA）,which was proposed by NVIDIA Corporation,is a development environment based on GPU general-purpose computation. In cognizance of the features of multi core of GPU,it could increase the speed of the massive data computing through parallel computation. In this paper,the CUDA based methods for parallel image sharpening,median filtering and character searching are proposed based on the CUDA technology and its features,and some key technologies and basic execution processes are discussed. The experimental results show that the computing speed of the proposed methods in most cases is much faster than that of the CPU based methods,but in some cases slower. It indicates that the CUDA technology has both merits and demerits and the experimental results provide references for complex applications.

关 键 词：图形处理器 统一计算设备构架 图像锐化 中值滤波 并行计算 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]