Research On Scheduling Algorithm Of Many-Core Processor Based On PSO And ACO

With the progress and development of semiconductor technology,the processor architecture has made great progress.Especially the emergence of multicore and manycore processor technology,the performance and efficiency of the processors have been greatly improved,which is an important support to improve the performance of computing systems,and reduce their power consumption.Because of the large number of processor cores on the CMP processors,how to further improve the utilization efficiency of the cores processor is an unavoidable problem,which is also an important problem to be solved.One of the key factors that determines the performance of the system is the scheduling of multiple tasks to the cores.At present,the researches on many core processor systems is relatively few,and most researches on single chip multiprocessor task scheduling is designed for multi-core processor systems.Multi-core processors and many core processors generally have a huge difference in the number of processors,so task scheduling algorithm for many core processor is more suitable to meet the requirements.In order to optimize the task scheduling in the system,a new task scheduling algorithm based on particle swarm optimization and ant colony algorithm is proposed in this dissertation.In this algorithm,all tasks are placed in a single core system at the initial stage,and a series of reasonable task scheduling sequences are generated;in the iteration process,A new pheromone updating method is added to avoid the rapid aggregation of particles;at the same time,through the introduction of genetic algorithm crossover and mutation strategy for local optimum and global optimum particle adaptive crossover and mutation,according to the fitness value of particle change update the location of the particle.The experimental results show that the improved algorithm is superior to the genetic algorithm in the performance of task scheduling.