Performance analysis of hybrid OpenMP/MPI based on multi-core cluster architecture

摘要

The mixed-mode OpenMP and MPI programming models in parallel application have significant impact on multi-core clusters in terms of performance. Hybrid programming has achieved new stage in parallel programming since the advent of new architecture of multi-core clusters. This new programming model leads ideas towards experimenting shared memory concept and message passing. The purpose of this paper is to analyze how pure MPI and hybrid code behave on two different multi-core clusters architecture using parallel application matrix multiplication. Hybrid programming code forced to face lack of performance compare to pure MPI at lower number of cores in overhead from shared memory. However, hybrid code outperforms pure MPI at higher counts of processors because of reduction in communication overhead that leads to lower process time.