计算机工程

• 体系结构与软件技术 • 上一篇    下一篇

面向通用计算GPU集群的任务自动分配系统

胡新明,盛冲冲,李佳佳,吴百锋   

  1. (复旦大学计算机科学技术学院,上海 201023)
  • 收稿日期:2013-02-27 出版日期:2014-03-15 发布日期:2014-03-13
  • 作者简介:胡新明(1989-),男,硕士研究生,主研方向:面向GPU的大规模并行计算;盛冲冲、李佳佳,硕士研究生;吴百锋,教授。

Automatic Task Assignment System of General Computing Oriented GPU Cluster

HU Xin-ming, SHENG Chong-chong, LI Jia-jia, WU Bai-feng   

  1. (School of Computer Science, Fudan University, Shanghai 201203, China)
  • Received:2013-02-27 Online:2014-03-15 Published:2014-03-13

摘要: 当前GPU集群的主流编程模型是MPI与CUDA的松散耦合,采用这种编程模型进行编程,存在编程复杂度大、程序的可移植性差、执行效率低等问题。为此,提出一种面向通用计算GPU集群的任务自动分配系统StreamMAP。对编译器进行改造,以编译制导的方式提供集群任务的计算资源需求,通过运行时系统动态地发现、建立并维护系统资源拓扑,设计一种较为契合GPU集群应用特征的任务分配策略。实验结果表明,StreamMAP系统能降低集群应用程序的编程复杂度,使之较为高效地利用GPU集群的计算资源,且程序的可移植性和可扩展性也得到了保证。

关键词: GPU集群, 异构, 编程模型, 任务分配, 可移植性, 可扩展性

Abstract: MPI+CUDA are the mainstream programming models of current GPU cluster architecture. However, by using such a low level programming model, programmers require detailed knowledge of the underlying architecture, which exerts a heavy burden. Besides, the program is less portability and inefficient. This paper proposes StreamMAP, an automatic task assignment system on GPU clusters. It provides powerful, yet concise language extension suitable to describe the compute resource demands of cluster tasks. It develops a run time system to maintain resource information, and supplies an automatic task assignment for GPU cluster. Experiments show that StreamMAP provides programmability, portability and scalability for GPU cluster application.

Key words: GPU cluster, heterogeneous, programming model, task assignment, portability, scalability

中图分类号: