CPU/GPU异构并行系统研究综述  

The Review of CPU/GPU Heterogeneous Parallel System

作　　者：屈宜丽[1] 蓝才会[1] 任志国[1] 

机构地区：[1]兰州城市学院信息工程学院,甘肃兰州730070

出　　处：《自动化与仪器仪表》2016年第4期25-26,共2页Automation & Instrumentation

摘　　要：近年来,CPU/GPU异构并行系统凭借其强大的计算能力、高性价比和低功耗等优良特点,成为深受各大领域欢迎的新型高性能计算平台,但其复杂的体系结构也为并行计算的研究提出了巨大的挑战,成为一个热门的新兴研究领域。本综述从CPU/GPU异构编程模型、存储模型、GPU集群、大数据环境下GPU系统的应用等几个方面对CPU/GPU异构并行系统的国内研究工作进行了回顾,并做了简要的分析,以期为初学者及其他研究者提供一定的参考。With the features of tremendous capability,high performance / price ratio and low power,the heterogeneous hybrid CPU /GPU parallel systems have become the new,popular high performance computing platforms. However,the architecture complexity of the hybrid system poses many challenges on the study of parallel computing,and thus become a new hot research field. In this paper,we reviewed domestic and foreign research work on the CPU / GPU heterogeneous from programming model,storage model and GPU cluster to the application in big data of GPU,and remark them briefly in order to provide some reference for beginners and other researchers.

关 键 词：CPU/GPU 异构并行系统 GPU集群 

分 类 号：TP338.6[自动化与计算机技术—计算机系统结构]