Longer Is Shorter: Making Long Paths to Improve the Worst-Case Response Time of DAG Tasks

Abstract

DAG (directed acyclic graph) tasks are widely used to model parallel real-time workload. The real-time performance of a DAG task not only depends on its total workload, but also its graph structure. Intuitively, with the same total workload, a DAG task with looser precedence constraints tends to have better real-time performance in terms of worst-case response time. However, this paper shows that actually we can shorten the worst-case response time of a DAG task by carefully adding new edges and constructing longer paths. We develop techniques based on the state-of-the-art DAG response time analysis techniques to properly add new edges so that the worst-case response time bound guaranteed by formal analysis can be significantly reduced. Experiments under different parameter settings demonstrate the effectiveness of the proposed techniques.