多核平台的并行实时调度与内存分配算法

多核处理器凭借着低功耗高性能的优势占据了市场.针对多核平台上并行实时任务,提出局部与全局EDF相结合的调度算法,其中任务的截止期划分、执行预算以及迁移时机由所设计的处理器带宽预留服务器决定.同时,提出了内存分配算法,该算法能够更好地为并行实时任务管理内存资源.实验结果表明新的调度算法具有更高的调度成功率.另外,在内存资源竞争的前提下,内存分配算法可以保证并行任务的实时性与系统稳定性.

Research on Parallel Real Time Scheduling and Memory Allocation Algorithm on Multi-core Platform

With low power consumption and high performance characteristics,multi-core processors have occupied the main market.Aiming at parallel real time scheduling on multi-core platform,a scheduling algorithm combined with local and global EDF was proposed.The budgets,deadline partition,and task migration time were decided by the proposed CPU width reserved server.A memory allocation method was presented,which could manage the memory resources for parallel real time tasks effectively.Experimental results showed that the proposed new scheduling algorithm has a higher scheduling success rate.In addition,using the presented memory partition algorithm,real-time characteristics and stability of tasks were assured in memory competition situation.