Abstract: This study aimed to measure the performance of an asynchronous algorithm of Particle Swarm Optimization. Particle Swarm Optimization (PSO) is a bio-inspired algorithm founded on the cooperative behavior of agents and is known as a tool to address difficult problems in numerous and divers fields. Like evolutionary algorithms, PSO offer practical approach to solve complex problems of realistic scale and gave results at least satisfactory. In addition, the performance of production systems is related to the scheduling of work on the one hand and to the assignment of this work of the various machines of the system on the other hand. The problem is noted Np-complete. Nevertheless, it remains that solving these problems require large computational demand in terms of CPU time and memory. Also, it is possible to improve solutions quality in various manners. In this study, we apply an asynchronous parallelization strategy of PSO algorithm on a scheduling problem in hybrid Flow-Shop (FSH) systems. We use a fault-tolerant environment by exploiting the computing power of a high-performance cluster with homogeneous processors. In a master/Slave model, PSO algorithm is decomposed to several tasks that are distributed on compute slave nodes. Experimental tests are compared with those obtained by the serial algorithm. Parallel performance is evaluated and improved PSO algorithm by accelerating convergence.