Hybrid Message-Passing and Shared-Memory Programming in a Molecular Dynamics Application On Multicore Clusters

Abstract

Hybrid programming, whereby shared-memory and message-passing programming techniques are combined within a single parallel application, has often been discussed as a method for increasing code performance on clusters of symmetric multiprocessors (SMPs). This paper examines whether the hybrid model brings any performance benefits for clusters based on multicore processors. A molecular dynamics application has been parallelized using both MPI and hybrid MPI/OpenMP programming models. The performance of this application has been examined on two high-end multicore clusters using both Infiniband and Gigabit Ethernet interconnects. The hybrid model has been found to perform well on the higher-latency Gigabit Ethernet connection, but offers no performance benefit on low-latency Infiniband interconnects. The changes in performance are attributed to the differing communication profiles of the hybrid and MPI codes.

Keywords

message passing shared memory multicore clusters hybrid programming

Get full access to this article

View all access options for this article.

References

Agarwal, A. 1992. Performance tradeoffs in multithreaded processors. IEEE Transactions on Parallel and Distributed Systems 3(5): 525-539. Available at http://www.scopus.com/scopus/inward/record.url?eid=2-s2.0-0026925622&partner=40&rel=R4.5.0.

Alam, S. , Agarwal, P. , Hampton, S. , Ong, H. and Vetter, J. 2008. Impact of multicores on large-scale molecular dynamics simulations. In IEEE International Symposium on Parallel and Distributed Processing, 2008 (IPDPS 2008), pp. 1-7.

Ashworth, M. , Bush, I. , Guest, M. , Sunderland, A. , Booth, S. , Hein, J. , Smith, L. , Stratford, K. and Curioni, A. 2005. HPCx: towards capability computing. Concurrency and Computation: Practice and Experience 17(10): 1329- 1361.

Cappello, F. and Etiemble, D. 2000. MPI versus MPI+ OpenMP on the IBM SP for the NAS Benchmarks . In Supercomputing, ACM/IEEE 2000 Conference, p. 12.

Car, R. and Parrinello, M. 1985. Unified approach for molecular dynamics and density-functional theory. Physical Review Letters 55(22): 2471-2474.

Dongarra, J. , Gannon, D. , Fox, G. and Kenned, K. 2007. The impact of multicore on computational science software . CTWatch Quarterly 3: 3-10.

Henty, D. 2000. Performance of hybrid message-passing and shared-memory parallelism for discrete element modelling. In Supercomputing, ACM/IEEE 2000 Conference, p. 10.

Hutter, J. and Curioni, A. 2005. Dual-level parallelism for ab initio molecular dynamics: reaching teraflop performance with the CPMD code. Parallel Computing 31(1): 1-17.

Jost, G. , Jin, H. , an Mey, D. and Hatay, F. 2003. Comparing the OpenMP, MPI, and hybrid programming paradigms on an SMP cluster. NASA Ames Research Center, Fifth European Workshop on OpenMP (EWOMP03), Aachen, Germany.

10.

Kazempour, V. , Fedorova, A. and Alagheband, P. 2008. Performance Implications of Cache Affinity on Multicore Processors (Lecture Notes in Computer Science, Vol. 5168), Springer, Berlin, pp. 151-161.

11.

Lusk, E. and Chan, A. 2008. Early Experiminents with the OpenMP/MPI Hybrid Programming Model (Lecture Notes in Computer Science, Vol. 5004), Springer, Berlin, p. 36.

12.

Message Passing Interface Forum 2008. MPI: A Message Passing Interface Standard Version 2.1, available at http://www.mpi-forum.org/.

13.

Nielsen, I. and Janssen, C. 2000. Multi-threading: a new dimension to massively parallel scientific computation. Computer Physics Communications 128(1): 238-244. Available at http://www.scopus.com/scopus/inward/record.url?eid=2-s2.0-0034625301&partner=40&rel=R4.5.0 .

14.

OpenMP Architecture Review Board 2007. OpenMP Application Programming Interface, Version 2.5, available at http://www.openmp.org.

15.

Rabenseifner, R. 2003. Hybrid parallel programming on HPC platforms. In Proceedings of the Fifth European Workshop on OpenMP, EWOMP, Vol. 3, pp. 22-26.

16.

Smith, L. and Bull, M. 2001. Development of mixed mode MPI/OpenMP applications. Scientific Programming 9: 83- 98. Available at: http://www.ingentaconnect.com/content/ios/spr/2001/00000009/F0020002/spr00078 .

17.

Strohmaier, E. , Dongarra, J. , Meuer, H. and Simon, H. 2005. Recent trends in the marketplace of high performance computing . Parallel Computing 31(3-4): 261-273. Availabel at: http://www.scopus.com/scopus/inward/record.url?eid=2-s2.0-18844445675&partner=40&rel=R4.5.0 .