基于计算量拟合的并行数据剖分算法  

Parallel data partition algorithm based on fitting of amount of computation

作　　者：陈宇澍[1] 宋君强[1] 吴建平[1] 龚西平[1] 刘志强[1] 

机构地区：[1]国防科学技术大学计算机学院,长沙410073

出　　处：《计算机工程与应用》2008年第10期34-36,39,共4页Computer Engineering and Applications

基　　金：国家自然科学基金(the National Natural Science Foundation of China under Grant No.40505023)

摘　　要：提出了一种基于计算量拟合的并行数据剖分算法,算法针对并行计算中静态负载不平衡问题,由各处理器实测计算时间,拟合出各基本数据剖分单元的计算量,在此基础上按各处理器计算时间相等的原则确定数据剖分。模拟试验表明,该算法能快速有效地实现各处理器的负载平衡。A data partition algorithm based on the fitting of the amount of computation is proposed.It estimates the amount of computation for each atomic data partition unit by numerical fitting from the computation time of each processor,and calculates the data partition following the principle that the computation time of each processor should be the same.h has been proved by experiments that the algorithm is efficient in load balancing.

关 键 词：数据剖分 负载平衡 基本数据剖分单元 计算量拟合 

分 类 号：TP302.7[自动化与计算机技术—计算机系统结构]