Research On Static Task Scheduling Strategy Based On Heterogeneous Multi-core Processors

From the coming of single-core processor, the research on task scheduling has been paidspecially attention of the exports and scholars not only for its direct impact on theperformance of the system but also the fact that the improvement of system performancerelies on both its own hardware level and software loaded on the hardware. With theemergence of multi-core processor, experts and scholars have turned their attention to the taskscheduling on multiprocessor, and the multiprocessor task scheduling has become one of theresearch hotspots in high performance processors. In order to meet the future developmenttrend of the processor, this paper is to research the task scheduling of the heterogeneousmulti-core processor, aiming at finding an ideal static task scheduling strategy to make thetask which running on the heterogeneous multi-core processors is scheduled reasonably, aswell as reducing the total length of scheduling, increasing the efficiency of multi-coreprocessors, and thus finally achieves the purpose of enhancing processor executionperformance.Based on the analysis and researches of the existing four classic scheduling algorithmson the heterogeneous multi-core, the paper proposes a static scheduling strategy based on theweighted priority, which overcomes the shortcomings existed in the current efficient taskscheduling algorithms, and the specific implementation method has been designed. Taskpriority values are calculated based on the orderly execution of task merge, task layering andtask weight calculation. The task mapping is implemented by the two alternate processes oftask distribution and optimization of task scheduling results. The new scheme ensures the taskis executed by the order of the layers and the critical tasks of each layer are priority scheduledto shorten the critical path length and reduce the total task completion time. In the schedulingprocess, the technology of task replication is used to advance the follow-up task executiontime by the execution of redundant task and the timely deletion of the redundant tasks whichcan shorten the time of task scheduling and improve the parallelism of the multi-coreprocessors system.In order to verify the efficiency and feasibility of this new kind of task schedulingstrategy based on weighted priority, the paper used experimental method to verify theproposed algorithm on high-performance platform Simics simulator though simulating themulti-core processor basic structure, and loading the algorithm in the simulator. Theexperimental results show that: compared with previous algorithms, the new algorithm has inherited the advantages of the critical task run priority, while the inappropriate choice of thepriority and the redundant task have been improved greatly, and the parallelism of the task inthe task processing is also improved, which reducing the total task scheduling length and theexpected effect of improved algorithm has been achieved. The new one has a certain practicalapplication value.