Dynamic multi-thread scheduling algorithm for better performance of the Alchemi toolkit

Grid computing refers to systems that harness the unused CPU cycles of personal computers connected over a corporate network or the Internet to accelerate application performance. There are some software toolkits used for implementing grid computing. Among them, Alchemi is widely used and open source toolkit that runs on the Windows operating system in the .NET Framework. The node which requests an application to be performed is called the owner. The node that receives the requested application and sends results back to the owner is called Manager. An application is divided into many threads. These threads are submitted to other nodes called Executor. In many cases, the faster one remains idle for a long time after completing the job. Current scheduling algorithm assigns only one thread to one executor. In this paper, we propose a technique to assign more than one thread to execute according the available CPU cycle.