集群环境下基于EasyHPS的并行程序调度算法的研究-计算机科学与技术专业论文.docxVIP

This paper then compares the HPS-FCSC scheduling algorithm with the original dynamic scheduling algorithm of EasyHPS by applying both on the Smith-Waterman algorithm (including both SWLAG and SWGG). This experiment has proved the validity of HPS-FCSC algorithm in the system of EasyHPS. Simulation has also been conducted in order to prove the effectiveness of HPS-FCSC algorithm in other types of application. In the simulation, four other list scheduling algorithms has been used in the parallel gaussian elimination programs as well as HPS-FCSC. The same experiment also conducted in other 20 randomly generated task graphs. Both results show the effectiveness of HPS-FCSC. KEY WORDS：parallel programming, scheduling algorithm, EasyHPS, bioinformatics 目 录 HYPERLINK \l _bookmark0 第一章 绪论1 HYPERLINK \l _bookmark1 1.1 研究背景1 HYPERLINK \l _bookmark2 1.2 国内外研究现状及发展趋势2 HYPERLINK \l _bookmark3 1.3 主要工作和贡献4 HYPERLINK \l _bookmark4 1.4 本文的内容和结构5 HYPERLINK \l _bookmark5 第二章 EasyHPS运行时系统概要 6 HYPERLINK \l _bookmark6 2.1 面向DAG数据驱动型应用的模型 6 HYPERLINK \l _bookmark7 2.1.1 模型的提出和意义6 HYPERLINK \l _bookmark9 2.1.2 模型框架7 HYPERLINK \l _bookmark12 2.2 任务调度与负载平衡9 HYPERLINK \l _bookmark13 2.3 容错机制9 HYPERLINK \l _bookmark16 2.4 数据流控制11 HYPERLINK \l _bookmark18 第三章 并行程序的调度算法理论13 HYPERLINK \l _bookmark19 3.1 调度问题的描述13 HYPERLINK \l _bookmark20 3.1.1 DAG模型 13 HYPERLINK \l _bookmark21 3.1.2 调度问题的定义13 HYPERLINK \l _bookmark22 3.1.3 如何衡量调度系统14 HYPERLINK \l _bookmark23 3.2 通信模型14 HYPERLINK \l _bookmark25 3.3 调度问题的复杂性16 HYPERLINK \l _bookmark26 3.3.1 不考虑通信时间的调度问题的复杂性17 HYPERLINK \l _bookmark27 3.3.2 考虑通信时间的调度问题的复杂性17 HYPERLINK \l _bookmark28 3.4 几类主要的调度算法18 HYPERLINK \l _bookmark30 3.4.1 表调度19 HYPERLINK \l _bookmark31 3.4.2 基于任务复制的调度19 HYPERLINK \l _bookmark33 3.4.3 基于任务集群或聚类的调度20 HYPERLINK \l _bookmark34 3.4.4 非确定性调度21 HYPERLINK \l _bookmark36 第四章 基于EasyHPS的HPS-FCSC调度算法 23 HYPERLINK \l _bookmark37 4.1 动态规划算法23 HYPERLINK \l _bookmark38 4.1.1 动态规划算法及其分类23 H