Predicting the runtime of tasks based on neural networks on grids
Application run-time information is a fundamental component in application and job scheduling. Predicting the runtime of a task, an important component of the resource management, plays an important role in the task scheduling and the resource using in computational grid. Such techniques can improve the performance of scheduling algorithms. However, the runtime of a task is a variable affected by many factors, accurate predictions of runtimes are difficult to achieve for parallel applications running in shared environments where resource capacities can change dynamically over time. This paper presents a predicting model for tasks runtime based on BP neural networks considering several factors. The method has many advantages including small network structure, quick learning and use conveniently etc. The result of prediction indicates that the method is effective and has higher accuracy.
computational grid predicting the runtime of tasks neural networks and BP algorithm
Jingbo Yuan Shunli Ding Jiubin Ju Liang Hu
College of Computer Science and Technology, Jilin University, Changchun, Jilin, China;Department of College of Computer Science and Technology, Jilin University, Changchun, Jilin, China
国际会议
杭州
英文
722-726
2006-10-12（万方平台首次上网日期，不代表论文的发表时间）