Research On A Hybrid Task Scheduling Algorithm Based On Heterogeneous Multi-core Processor

With the development of computer architecture,the internal structure of multi-core processor is more and more complex,and the number and type of the processor cores are increasing.How to make full use of these processor cores has become a hot topic in the field of computer science.In the multi-core processor,the task scheduling mechanism directly affects the utilization of the cores,only the appropriate task scheduling algorithm can play the multi-core processor high performance advantages.Some existing task scheduling algorithms have problems of low scheduling efficiency and poor scheduling result quality.In order to further improve the parallel processing capability of multi-core processors,it is of great significance to design an efficient task scheduling algorithm.In this paper,the existing multi-core processor architecture,task scheduling system and task scheduling algorithm are studied in depth.Combining the advantages of list scheduling algorithm and genetic algorithm,focusing on how to overcome the shortcomings of the two algorithms,the task scheduling algorithm ILSGA is proposed.ILSGA mainly contains two parts.The first part is an improved list scheduling algorithm called ICPOP,which focuses on improving task priority calculation method and processor cores allocation strategy,which can generate a better task scheduling sequence.The second part is optimizing the genetic algorithm.The idea of optimization is to form a chromosome in the initialized population by using the optimal task scheduling sequence generated by ICPOP algorithm,the remaining chromosomes are randomly generated according to the height value of the task in the task graph,thus a better initial population shortens the convergence time of the algorithm.Then the optimal chromosomes are saved by combining the tournament selection strategy with the dynamic selection strategy based on the fitness value,so as to avoid the local convergence of the algorithm.Finally,the crossover and mutation of chromosomes in the population are carried out by adaptive crossover and mutation probability to avoid "premature" phenomenon,thus,obtaining the optimal task scheduling sequence.In order to verify the availability and efficiency of ILSGA in the heterogeneous multi-core processors task scheduling system,the experiments randomly generate different DAGs as the input cases,and the Matlab2010 b programming environment is used to analyze and verify the algorithm.The experimental results show that the ILSGA has the shortest task scheduling length on the same conditions,compared with CPOP and GA,and has a good convergence rate.ILSGA has a certain availability,stability and efficiency in the task scheduling system.