Research On IGSA Task Scheduling Algorithm Based On Chip Multi-processors

With the advent of the 5G era and the rapid development of the modern Internet,the rapid increase in user data has ushered in a golden period of computer architecture.Multi-core architecture has become one of the hotspots of modern mainstream processor research.Processor,CMP),also known as multi-core microprocessor,has become one of the bridges connecting the present and the future digital economy.The field of CMP is booming and contains endless scientific research,economic and commercial potential.The organic combination of CMP and task scheduling algorithms will have an immeasurable impact on the development of computers.Therefore,it is sufficient to study an efficient,fast and stable task scheduling algorithm.It is imperative to give full play to the performance of CMP.This paper proposes an improved algorithm based on simulated annealing algorithm and applies it to the field of CMP task scheduling.The improvement strategies are as follows: first,the concept of fitness value variation coefficient is introduced to represent the differences between individuals in the population.In the initialization stage,the parallelization method based on variation coefficient is adopted to expand the scope of CMP task scheduling The range of the initial solution set in the state space makes the distribution of individual solutions in the initial solution set more dispersed and uniform;secondly,the explosion operator of fireworks algorithm is introduced and its explosion range is improved to replace the random disturbance strategy in the original algorithm to control the range of children in the state space and enhance the search ability in the local space,reduce the number of iterations and improve the convergence speed of the algorithm;thirdly,the algorithm is improved The simulated annealing algorithm only depends on the disturbance function to generate the offspring,which is not mutagenic,so the gene mutation strategy is added to supplement the process,so that the generated offspring have diversity,so as to enrich the types of the offspring population and improve the diversity.By establishing a dependent task model,designing a reasonable encoding and decoding method and fitness value function,the continuous solution space of IGSA algorithm is mapped to the discrete space of CMP task scheduling problem to obtain high-quality task scheduling sequence.In order to test the performance of IGSA algorithm in dealing with heterogeneous CMP static dependent task scheduling problem,comparative experiments are carried out on the MATLAB simulation platform.The IGSA task scheduling algorithm is compared with simulated annealing algorithm,fireworks algorithm and genetic algorithm respectively.The optimal task scheduling sequence length and iteration number under different system environment and different number of tasks are obtained and analyzed the statistical analysis and performance evaluation are given.The simulation results show that IGSA algorithm has good optimization ability in heterogeneous CMP static dependent task scheduling problem,less iterations,and higher quality task scheduling sequence in a shorter time,which has a wide application prospect.