摘要
分析了当前的DAG调度算法,针对DAG调度中个别大任务制约任务完成时间的情况,提出了关键任务和资源预留时机的定义以及动态资源预留的思想,并基于动态资源预留的思想提出了一种网格环境中基于DAG的任务映射算法KtRR。测试结果表明,该算法比已有算法效率高,特别是在任务密集的情况下。
By analyzing the existing DAG scheduling algorithms, and focusing on the fact that some large subtasks heavily undermine the efficiency in completing tasks in DAG scheduling, the definitions of key task and resource reserving opportunity, as well as the idea of dynamically reserving resource for key task are brought forward. Furthermore, a task-mapping algorithm based on DAG in grid env ironment: KtRR is also provided on the basis of dynamical resource reservation. R esults show that the KtRR is more efficient than current ones, especially in task -dense situations.
出处
《计算机应用研究》
CSCD
北大核心
2005年第7期12-13,56,共3页
Application Research of Computers
基金
国家自然科学基金重点项目(59835170)
中南大学博士后基金资助项目
