论文部分内容阅读
资源预留是实时系统中一种实现任务间时间隔离的方法。实时程序在多核处理器上以资源预留的方式运行,可将其划分为多个执行流,并将各执行流部署到处理器核上。为了节省计算资源占用和减少处理器能耗,在部署时应该使用尽量少的处理器核。本文将该问题映射为装箱问题(Bin packing),并提出基于首次适应算法(FF)的FF—SA算法进行求解。FF—SA算法首先利用首次适应算法(FF)得到执行流部署的初始解,其次再利用模拟退火算法(SA)进一步优化。实验结果表明,FF—SA算法比FF算法节省处理器核数。
Resource reservation is a real-time system to achieve a method of time separation between tasks. The real-time program runs on a multi-core processor in a resource reservation manner, divides it into multiple execution streams, and deploys each execution stream to the processor core. To save computing resources and reduce processor power consumption, you should use as few processor cores as possible in your deployment. In this paper, the problem is mapped to Bin packing, and the FF-SA algorithm based on the first adaptive algorithm (FF) is proposed to solve the problem. The FF-SA algorithm first uses the first-time adaptive algorithm (FF) to obtain the initial solution of the execution flow deployment, and secondly, it further optimizes by the simulated annealing algorithm (SA). Experimental results show that the FF-SA algorithm saves processor cores compared to the FF algorithm.