GPGPU技术研究与发展  

Research and Development of the General-Purpose Computation on GPUs

作　　者：林一松[1] 唐玉华[1] 唐滔[1] 

机构地区：[1]国防科学技术大学计算机学院,湖南长沙410073

出　　处：《计算机工程与科学》2011年第10期85-92,共8页Computer Engineering & Science

基　　金：国家自然科学基金资助项目(60621003;60873016)

摘　　要：半导体工艺的发展使得芯片上集成的晶体管数目不断增加,图形处理器的存储和计算能力也越来越强大。目前,GPU的峰值运算能力已经远远超出主流的CPU,它在非图形计算领域,特别是高性能计算领域的潜力已经引起越来越多研究者的关注。本文介绍了GPU用于通用计算的原理以及目前学术界和产业界关于GPGPU体系结构和编程模型方面的最新研究成果。With the development of the semiconductor technology, the number of transistors integrated on a chip keeps increasing. Consequently, the computation and memory capacity of graphics processing units improve rapidly. So far, the floating-point computing capacity of GPUs has greatly exceeded that of CPUs, and the potential of GPUs in the non-graphic computing field, especially in high per- formance computing, has attracted more and more researchers＇ attention. This paper gives an introduction to the principles of the general purpose computation on GPUs and the latest research results about architecture and the programming model of GPGPU from both the research community and industry

关 键 词：GPGPU 体系结构 编程模型 

分 类 号：TP303[自动化与计算机技术—计算机系统结构]