Task Scheduling in Multiprocessor System using Fork-Join Method (TSFJ)

The performance of the multiprocessor system depends on how the tasks are scheduled in multiprocessors. If we allocate tasks wisely then we can get good performance of the system. The primary goal of task scheduling in multiprocessor system is to minimize the total execution time, so that we can achieve maximum speed-up and efficiency. The application program is represented by directed acyclic graph (DAG) in scheduling problem. In this paper, we have proposed an algorithm for task scheduling in multiprocessor system based on forkjoin method. Tasks are allocated to the processors follows fork-join structure. The performance of the TSFJ algorithm is measured based on scheduling length, speedup and efficiency. Further, we have shown the comparative study between TSFJ algorithm and heuristic algorithms.