基于GPU的异构并行编程模型分析与研究  

Analysis and Research of Heterogeneous Parallel Programming Model Based on GPU

作　　者：包达尔罕 高文炜 郑欣 冯路 杨金颖 BAO Daerhan;GAO Wen-wei;ZHENG Xin;FENG Lu;YANG Jin-Ying(Xian Microelectronice Technology Institute,Xian 710054,China;Rocket Forces Equipment Department in Xian No.4 Representative Room,Xian 710054,China)

机构地区：[1]西安微电子技术研究所,陕西西安710054 [2]火箭军装备部驻西安地区第四代表室,陕西西安710054

出　　处：《内蒙古民族大学学报（自然科学版）》2020年第1期42-46,共5页Journal of Inner Mongolia Minzu University：Natural Sciences

摘　　要：近年来,异构系统硬件飞速发展.为了解决相应的编程和执行效率问题,异构并行编程模型已经被广泛使用和研究.从异构并行编程模型与编程框架两个角度总结了异构并行编程模型的最新研究成果.异构并行编程为异构系统与上层应用之间搭建桥梁,为当前高速发展的并行计算提供了技术方案支持.在主流的并行编程模型中,CUDA与OpenCL是广泛被人们所青睐的两种方案,并且它们能够提供成熟且稳定的技术支持.文章以CUDA和OpenCL的特征为出发点,研究了异构并行编程模型的核心技术.To solve the corresponding problems of programming and execution efficiency,heterogeneous parallel programming model has been widely used and studied.The latest research results on heterogeneous parallel programming model and compilation structure,are summarized.Heterogeneous parallel programming builds a bridge between the heterogeneous system and the upper application,which provides technical support for the current high-speed parallel computing CUDA and OpenCL,as two widely accepted solutions in the mainstream parallel programming model,providing mature and stable technical support.Based on the programming model and framework between the two programming languages,the core technologies of the heterogeneous parallel programming model are discussed in depth.Finally,the future direction is proposed based on the current research status and the development of the heterogeneous system.

关 键 词：CUDA OPENCL 编程模型 编程框架 

分 类 号：TP332[自动化与计算机技术—计算机系统结构]