Research On Task Scheduling Of Pseudo Dynamic Based On Heterogeneous Multi-core Processors

At present,Single-core processors have cannot meet the requirements of processing capabilities for its organizational structure and working methods. Multi-core processors has became the new direction of development processor.The Research to improve its Performance is more and more important Task scheduling to be the most important of the processor resource allocation.This paper expand research to the problem of the task scheduling algorithm that critical path select imprecise and the first task scheduling efficiency is not high. Proposed static scheduling algorithm based on heterogeneous multi-core processors called pseudo dynamic task scheduling algorithm. This algorithm includs three stages.First at the priority determining stage, Determine the task graph collection that is each processor cores specific.Select the longest critical path to be the current critical path. When calculate the priority that give the higest priority to the critical task and use the manner of calculation of weights to prioritize other tasks; Secondly at the task mapping stage, select the smallest EFT of the processor to be target scheduling core. Through the way of the local optimum to complete schedule within the optimal time. Improve the efficiency of the processor core scheduling; At last, at the status update stage this paper gives two update guidelines:task compute overhead and communication overhead. At the same time,at each step is completed in real-time scheduling status updates, Ensure the accuracy of the calculation of the current critical path. This ensures that the urgency of the task of high priority to schedule. Makeing the task of scheduling efficiency and parallel and compute power of multicore processors get promoted.Research on task scheduling of pseudo dynamic critical path based on heterogeneous multi-core processor at simics experimental platform to Verificate. In order to ensure the fairness of algorithm, define three experimental parameters as a measure of algorithm performance standards. The method uses randomly generated DAG to obtain the test set.Obtain three sets of experiments to performance comparison. Experimental result shows that this paper's algorithm overcome the shortcoming that existing algorithms selecte critical paths improperly and shorten the length of task scheduling and improve the performance of task scheduling...