2011 | OriginalPaper | Chapter
Analysis of Core-Level Scale-Out Efficiency for OpenMP Programs on Multi-core Processors
Authors : Sareh Doroodian, Nima Ghaemian, Mohsen Sharifi
Published in: High Performance Architecture and Grid Computing
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
The majority of existing OpenMP compilers select the maximum number of available processing cores on a multi-core machine at runtime to execute a parallelized program on that machine. In this paper, we show that the use of maximum number of available cores does not necessarily result in speedup or efficiency. We show that in a considerable number of cases the use of more cores results in diminishing returns on execution time and efficiency. To help in choosing the proper number of cores, we propose an analytical method to estimate the execution times of OpenMP programs using different numbers of cores while considering the synchronization, excess computation and load imbalance overheads caused by the chosen number of cores. We validate our proposed method through a case study covering most recurrent and important structures and constructs of OpenMP.