首页 | 本学科首页   官方微博 | 高级检索  
     

基于多核集群系统的并行编程模型的研究
引用本文:胡晨骏 王晓蔚. 基于多核集群系统的并行编程模型的研究[J]. 微机发展, 2008, 18(4): 70-73
作者姓名:胡晨骏 王晓蔚
作者单位:东南大学计算机科学与工程学院 江苏南京210096
摘    要:并行计算技术是计算机技术发展的重要方向之一。当前并行程序模型主要有消息传递模型和共享存储模型两种。随着处理器多核技术的发展,在一枚多核处理器中集成两个或多个完整的计算引擎(内核),并充分利用多核计算机的特性,发挥多核计算机的性能成为一个很重要的研究方向。介绍一种新的MPI实现机制,这种机制集成了共享存储模型和消息通信模型的优点,在节点内使用共享存储模型,在节点间使用消息传递模型,并且通过自动生成线程级的任务来获得更好的性能。.

关 键 词:并行编程模型  消息传递模型  共享存储模型  MPI  线程
文章编号:1673-629X(2008)04-0070-04
修稿时间:2007-07-11

Research of Parallel Programming Model Based on Multi-Core Cluster System
HU Chen-jun,WANG Xiao-wei. Research of Parallel Programming Model Based on Multi-Core Cluster System[J]. Microcomputer Development, 2008, 18(4): 70-73
Authors:HU Chen-jun  WANG Xiao-wei
Affiliation:HU Chen-jun, WANG Xiao-wei (School of Computer Science and Technology,Southeast University, Nanjing 210096, China)
Abstract:Parallel computing is one of the most important techniques of computing.Contemporary parallel programming model can be roughly divided into message-passing model and shared-memory model.With the development of multi-core processor,this integrated two or more intact computing engine(kernel) in one processor.There arouses a very important researching direction as how to effectively utilize the feature of the multi-core to achieve optimal performance.Introduce a new implementation of MPI,which could integrates the features of both the shared-memory model and the message-passing model,that is,use shared-memory model within one node while use message-passing model among nodes.And more over,better performance is achieved through automatic thread-level task spawning mechanism.
Keywords:parallel programming model  message-passing model  shared-memory model  MPI  thread
本文献已被 CNKI 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号