In heterogeneous, distributed and shared grid environment, a way of gaining high performance is to design a good grid scheduler. But a successful scheduling mainly depends on both system information and application sp...In heterogeneous, distributed and shared grid environment, a way of gaining high performance is to design a good grid scheduler. But a successful scheduling mainly depends on both system information and application specific information. To obtain more information from application, this paper studies the intensive characteristics of grid computing task. A novel algorithm for recognizing the style of intensive program is proposed. Hence an application task can be mapped onto an appropriate grid resource, and run very well.展开更多
基金This work is supported by the 863 High-Tech Project (No. 2004AA104340), the National Natural Science Foundation of China (No. 60173026) and SEC E-Institute: Shanghai High Institutions Grid (No. 200301-1).
摘要In heterogeneous, distributed and shared grid environment, a way of gaining high performance is to design a good grid scheduler. But a successful scheduling mainly depends on both system information and application specific information. To obtain more information from application, this paper studies the intensive characteristics of grid computing task. A novel algorithm for recognizing the style of intensive program is proposed. Hence an application task can be mapped onto an appropriate grid resource, and run very well.