Research On Task Scheduling For Heterogeneous Multi-core Processors

The multi-core processors have become the mainstream of processor.With multiple processing cores,the application can make parallel threads simultaneously delivered to multiple cores respectively,thus the running speed of the program has been greatly improved.According to the differences in the kernel structure,multi-processors can be divided into homogeneous multi-core processor and heterogeneous multi-core processor.Heterogeneous multi-core select the suitable sub tasks according to the characteristics of the tasks,which makes the overall system operation performance significantly improved.The development of heterogeneous multi-core processor for study of task scheduling is to provide a broad space for development and the performance of the task scheduling algorithm directly affects the performance of the system.The main work of the paper is as follows:An energy aware algorithm based on deleting the redundant replicate tasks is proposed.The idea of the strategy is to execute the task in multi-processors so as to reducing the communication between tasks,which could reduce the earliest completion time and accelerated the execution speed.Although the strategy of copy could bring the improvement of the performance,it bring more energy consumption.In the scheduling process of the application,the overall schedule length is related with the execution time of critical tasks,a slight delay of the non-critical tasks does not affect the overall schedule length of the application.In this paper,our strategy is that if the process of deleting the replication tasks does not affect the relationship between tasks and the execution of the next task the processor will executed next time,the replication task could be deleted.The experiment results show that the proposed algorithm could effectively reduce the numbers of duplication task and reduce the energy consumption of the system.An improved list scheduling algorithm based on task replication is proposed.In the calculating the priority,the parameter is simplex in the existing list scheduling algorithm,which only consider the execution time and the communication from the current task to the exit task,but in the heterogeneous environment,the execution time of task is different on processors,the execution time may be more different when a task is assigned to different processor.So we improve the method of calculating the task priority,which fully consider the dependency of the task to the processor.In the stage of the task scheduling,some task could be executed even if the data from its immediate parents has arrived.This paper improves the scheduling process.The first is the stratification of DAG,the tasks in the same level is independent,so they could be executed in parallel.In the end of scheduling of a task,we search the other tasks in the same layer,if its data has arrived,we schedule the task in the idle time of processor.The experiments shows that,the performance of the algorithm proposed in this paper can effectively enhance the scheduling.