Comparative and Combined Performance Studies of OpenMP and MPI Codes

OpenMP와 MPI 코드의 상대적, 혼합적 성능 고찰

  • 이명호 (명지대학교 컴퓨터소프트웨어학과)
  • Published : 2006.04.01


Recent High Performance Computing (HPC) platforms can be classified as Shared-Memory Multiprocessors (SMP), Massively Parallel Processors (MPP), and Clusters of computing nodes. These platforms are deployed in many scientific and engineering applications which require very high demand on computing power. In order to realize an optimal performance for these applications, it is crucial to find and use the suitable computing platforms and programming paradigms. In this paper, we use SPEC HPC 2002 benchmark suite developed in various parallel programming models (MPI, OpenMP, and hybrid of MPI/OpenMP) to find an optimal computing environments and programming paradigms for them through their performance analyses.


  1. Rocco Aversa, Beniamino Di Martino, Nicola Mazzoca, Salvatore Venticinque, 'Performance Analysis of Hybrid OpenMP/MPI N-Body Application,' 5th International Workshop on OpenMP Applications and Tools, Houston, TX, May, 2004
  2. Manojkumar Krishnan, Yuri Alexeev, Theresa L. Windus, Jarek Nieplocha, 'Multilevel Parallelism in Computational Chemistry using Common Component Architecture and Global Arrays,' Super Computing (SC) 05, Seattle, WA, November, 2005
  3. Myungho Lee, Larry Meadows, Darryl Gave, Dominic Paulraj, Sanjay Gail, Brian Whitney, Nawal Copty, and Yonghong Song, 'Compiler Support and Performance Tuning of OpenMP Programs on SunFire Servers,' European Workshop on OpenMP, Aachen, Germany, September, 2003
  4. Message Passing Interface Standard,
  5. OpenMP Architecture Review Board,
  6. SPEC HPC2002 benchmark suite,
  7. Sun Fire 6800 server,
  8. Sun Fire 15K server,
  9. Sun Fire V880 server,
  10. Solaris 9 Operating System,
  11. Sun Studio 9 Compiler Suite,