Research On Task Scheduling Algorithms Based On Heterogeneous Multicore Processors

In the rapidly developing information age,computers have been in the state of iterative update,processor architecture is becoming more and more complex,and how to improve the utilization of processor cores,task scheduling mechanism is one of the keys,and heterogeneous multi-core processor is the key target of research today.This paper is mainly focused upon the scheduling of jobs in heterogeneous multicore processors,using task scheduling algorithms to reasonably schedule tasks in order to achieve optimal performance of heterogeneous multicore processors.In an effort to maximize the utilization of heterogeneous multi-core processors and improve the job scheduling rate,this article introduces a comprehensive task dispatch method by improving the existing table scheduling algorithm and introducing the lion swarm optimization algorithm.The main work is as follows:(1)To address the inefficiency of task scheduling caused by the Heterogeneous Critical Node First algorithm in a heterogeneous multicore environment due to the generation of redundant tasks and the failure to adequately consider the ordering of non-critical tasks,this paper proposes an optimized table scheduling algorithm Improved Heterogeneous Critical Node First.The algorithm first improves the task average execution overhead parameter on the weight calculation formula to task standard deviation execution overhead parameter in the task ordering stage,which can fully reflect the variability of tasks on heterogeneous multi-core processor cores,and tasks with large standard deviation are prioritized for scheduling,while also introducing the maximum precursor task communication overhead and maximum successor task communication overhead,and tasks with large task communication overhead are prioritized for scheduling,and the priority ranking of non-critical tasks is determined by the above three parameters together.Then the redundant task optimization strategy is used in the task assignment stage,firstly,the redundant tasks are judged separately for each layer of DAG tasks,and if there is no impact on the succeeding tasks when the task is removed,then the redundant mark is made and the redundant task is cleared.The results show that the IHCNF algorithm performs better compared with other algorithms and effectively improves the scheduling efficiency.(2)In order to further improve the scheduling efficiency and scheduling accuracy for the task scheduling problem,this paper will introduce the lion swarm optimization algorithm and propose a comprehensive task scheduling algorithm based on heterogeneous multi-core.Firstly,in order to optimize the lion swarm algorithm,the quality of the solution is improved by adding the Tent chaos mapping operator in the initialization stage,and the update formula of the lion cub is improved by introducing of the adjustment factor improves the update formula for cubs,increases the speed of solution and avoids getting trapped in a local optimum solution;The improved lion-group algorithm is then discretized to take the form of a prioritized encoding such that each individual lion-group position is mapped to a sequence of task scheduling;The combination of the two algorithms lies in the existence of the excellent scheduling sequence already obtained by the IHCNF algorithm as an individual of the initial population of the improved lion swarm algorithm,and then using the iterative merit-seeking of the lion swarm algorithm to set up the maximum number of iterations and finally output the optimal scheduling sequence.This comprehensive algorithm further improves the scheduling efficiency of the task.Finally,the proposed scheduling algorithm is tested comprehensively in a heterogeneous multicore system environment.Firstly,the corresponding experimental evaluation functions are set up and several test task models are randomly generated.After comparing with the lion swarm scheduling algorithm and the typical genetic algorithm,we can see that the algorithm in this paper is better in scheduling lower bound ratio and acceleration,and its scheduling performance is more stable and efficient.