摘要
文章针对具有复杂依赖关系的实时周期任务难以调度的问题,提出了一种模型转化方法,该方法将具有复杂关系的实时结点任务图转化为并行优先级任务树,然后根据模型中定义的优先关系并结合EDF(Earliest Deadline First)算法,将任务调度到相应的核上去执行。仿真试验表明,该方法比核与任务相对应的分配方法不仅可以提高CPU利用率而且大大减少了核间通信的开销。
Sections 1 through 4 of the full paper explain and evaluate the improvement mentioned in the title. The core of our thinking and that of sections 1 through 4 consists of： （ 1 ） past research papers on the multi-core scheduling model only focus on the independent tasks or the dependent tasks that are not real-time tasks; we propose a new scheduling model PPTT which can study the mutually dependent real-time tasks; it can ensure the satisfaction of the dependence relationship between any two tasks and can work together with the EDF（ Earliest Deadline First） algorithm to schedule these real-time tasks running on the corresponding cores; （2） section 1 is entitled ＂Scheduling Algorithm Based on PFFF Mathematical Model＂ ; （3） section 2 is entitled ＂Establishing Priority Relationship＂ ; （4） section 3 in entitled ＂Allocating Algorithm of Multiprocessor＂ ; （5） simulation results, presented in Fig. 4, and their analysis show preliminarily that the model and algorithm presented in this paper are indeed effective and the core has indeed a higher utilization.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2012年第5期652-656,共5页
Journal of Northwestern Polytechnical University
基金
航空科学基金(20100753022)
航天科技创新基金(2011XR160001)
西北工业大学基础研究基金(JC201102837)资助
关键词
并行优先级任务树
有向无环图
多核
实时任务
algorithms, computer simulation, control, design, efficiency, embedded systems, flowcharting, math ematical models, muhiprocessing systems, multitasking, real time systems, resource allocation, scheduling, synchronization
directed acyclic graph （DAG）, multi-core, parallel priority task tree（PPTT）
