Estimating overheads of OpenMP directives

Estimating the execution time of programs has always been a concern in computer science. With the emergence of multi-core processors, this concern has found new perspectives and new parameters affect the runtime performance of parallel applications. To estimate the execution time of parallel applications, we investigate the overheads caused by parallelizing an application by identifying the overheads of OpenMP directives. We show that the overheads caused by the implementation of synchronization directives are negligible.