Skip to Main Content
Energy consumption is a critical issue in parallel and distributed systems. Energy-efficient scheduling of directed acyclic graph (DAG) based workflows on dynamic voltage scaling (DVS) enabled systems consists of two phases: assignment of tasks and slack allocation. Most current research on scheduling for energy minimization of DAGs tries to minimize energy by effective slack allocation for a given assignment. The assignment itself does not take energy profiles of tasks into account. In this paper, we show that incorporating DVS based energy profiles of tasks during the assignment phase can lead to significantly lower overall energy requirements while requiring lower computational time.