the mapping of tasks among processors, which is known as task scheduling is one of the most important issues in multi-processor system. Although various solutions have been proposed, the most optimum method has not been found yet.Therefore, some expert in this field have tried developmental algorithms such as PSO and genetic. In this article, we try to study this problem by using the PSO algorithm. However, since this is an MPHard problem and the above mentioned algorithms are of high time complexity, we have tried to increase the speed of obtaining answers, without any resultant decrease in efficiency, through making changes in the PSO algorithm. Results presented in the experimental result of this article show that the speed of convergence increases as compared to the standard PSO.