A Study of enhancing statistic-computing performance on heterogeneous computing environments

• Main Authors: Pei-Shan Sung, 宋佩珊
• Other Authors: Guan-Joe Lai
• Format: Others
• Language: en_US
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/74531768794696494752

碩士 === 臺中師範學院 === 教育測驗統計研究所 === 91 === The heterogeneous computing system could exploit the computational powers of the tasks in an application. It is the key to parallelize the tasks to several processors to reduce the total execution time. Since HC environments could meet the requirement of exploiting the computational powers, so the HC environment is studied in this thesis. As a result, in this study, we are concerned about exploiting a competent list-based heuristic algorithm, which is called the Dominant Tasks Scheduling (DTS) algorithm, for scheduling the tasks of a parallel application into HC environments. In the systems with high communication heterogeneity or high computation heterogeneity, the DTS algorithm, could perform better than other proposed algorithms from the literature by considering global scheduling information and exploiting schedule-holes. The experimental results show the superiority of the DTS algorithm.