基于JXTA的点对点计算框架  

JXTA Based Computing Framework

作　　者：张国春[1] 张建军[1] 

机构地区：[1]西北大学信息科学与技术学院,陕西西安710127

出　　处：《微计算机信息》2008年第33期254-256,共3页Control & Automation

摘　　要：传统意义上的并行计算一般都是指细粒度的并行计算。近年来,逐渐出现了以粗粒度方式进行高效并行计算的需求。此类计算涉及诸如计算的吞吐量,执行大量相似但独立的任务以解决大规模问题;或依赖总体负载平衡的解决方案,在不同初始条件运行模拟运算,然后结果进行合并。本文展示了一个解决大规模、粗粒度并行计算的总体框架。这个框架基于具有广泛平台和组件支持Java语言,以及由JXTA协议提供的点对点通信机制,允许动态和非集中地管理可计算资源。Parallel computation has been an essential component of scientic computing for decades. Traditionally, when one thinks of parallelization, one often envisions fine-grained parallelization. Recently, however, there is an increasing demand for efficient mecha- nisms of carrying out computations which exhibit coarse-grained parallelism. Examples of this class of problems include throughput computations, where numerous similar but independent tasks are performed to solve a large problem, or any solution which relies on ensemble averages, where a simulation is run under a variety of initial conditions which are then combined to form the result. This paper presents a framework for these types of computations. The framework in this paper utilized the JXTA open peer-to-peer communication protocols, which allow for the dynamic aspect of the grid through peer discovery, in addition to the scalability aspect through the use of peer groups.

关 键 词：点对点 网格 JXTA 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]