| Authors | Z. Vrba, H. Espeland, P. Halvorsen and C. Griwodz |
| Editors | E. Frachtenberg and U. Schwiegelshohn |
| Title | Limits of Work-Stealing Scheduling |
| Afilliation | Communication Systems, Communication Systems |
| Status | Published |
| Publication Type | Proceedings, refereed |
| Year of Publication | 2009 |
| Conference Name | Job Scheduling Strategies for Parallel Processing (14th International Workshop, 2009) |
| Pagination | 280-299 |
| Publisher | Springer Berlin / Heidelberg |
| ISBN Number | 978-3-642-04632-2 |
| Abstract | The number of applications with many parallel cooperating processes is steadily increasing, and developing efficient runtimes for their execution is an important task. Several frameworks have been developed, such as MapReduce and Dryad, but developing scheduling mechanisms that take into account processing \emph{and} communication requirements is hard. In this paper, we explore the limits of work stealing scheduler, which has empirically been shown to perform well, and evaluate load-balancing based on graph partitioning as an orthogonal approach. All the algorithms are implemented in our Nornir runtime system, and our experiments on a multi-core workstation machine show that the main cause of performance degradation of work stealing is when very little processing time, which we quantify exactly, is performed per message. This is the type of workload in which graph partitioning has the potential to achieve better performance than work-stealing. |
| DOI | 10.1007/978-3-642-04633-9\_15 |
| Citation Key | Simula.ND.325 |