An Algorithm for Assigning Tasks in Parallel Computing
Yang Rui, Xinyu Zhang · 2010
As the parallel distributed architectures are increasing used to solve the large-scale scientific problems, the scheduling algorithm which can heavily affect the communication overhead and the execution time is becoming more and more important for the optimization of parallel programs in real distribute system or in the performance prediction to make more precise, so it requires appropriate techniques to identify the factors which influence the application execution time and also the way they interact. Generally, parallel applications characterized as a collection of tasks and their interrelations. As the number of the processes increasing, the overhead of the communication among these processes can be very impressive, different tasks assignment algorithm can affect the performance of the parallel application. This paper introduces a scheduling algorithm TMDC (Task Mapping based on Dependence and Communication) for assigning tasks to the processors and minimizes the communications volume to optimize the total execution time. It's based on the DAGs can helps to understand the details of the application and to increase the accuracy of the prediction models.