Profiling of task-based applications on shared memory machines: Scalability and bottlenecks

Abstract. A sophisticated approach for the parallel execution of irreg-ular applications on parallel shared memory machines is the decompo-sition into fine-grained tasks. These tasks can be executed using a task pool which handles the scheduling of the tasks independently of the ap-plication. In this paper we present a transparent way to profile irregular applications using task pools without modifying the source code of the application. We show that it is possible to identify critical tasks which prevent scalability and to locate bottlenecks inside the application. We show that the profiling information can be used to determine a coarse estimation of the execution time for a given number of processors.