Improved DAG Tasks Stretching Algorithm Based on Multi-core Processors

The performance of multi-core real-time operating system is greatly affected by the quality of tasks scheduling algorithm. Directed acyclic graph (DAG) task model can better represent the dependency between subtasks in multi-core real-time system. However, the scheduling algorithm previously applied to independent sequential tasks is no longer suitable for this kind of tasks, the efficient DAG tasks scheduling algorithm has become the current research hotspot. This paper discusses the stretching algorithm with excellent performance, and on the basis of the stretching algorithm, an improved stretching algorithm is proposed to overcome its shortcoming of ignoring dependence between subtasks. The performance optimization is verified by simulation experiments.