Uma Abordagem para Composição de Clusters Eficientes na Execução do Modelo Numérico WRF de Previsão do Tempo
Resumo
A resolução de problemas conhecidos por grand challenge, como é o caso da previsão do tempo por meio de modelos numéricos, demandam computação de alto desempenho. Apesar da consolidação dos clusters como solução para prover alto desempenho, a escolha dos computadores que o compõe está submetida à variabilidade das configurações disponíveis no mercado. De fato, a inserção de processadores multi-core em ambientes de cluster cria um cenário distinto no que diz respeito à comunicação entre processos. Nesse contexto, propõe-se uma abordagem em que alguns núcleos de processamento não são alocados a processos da aplicação, com o intuito de construir clusters econômicos mas também eficientes, interconectados por Gigabit Ethernet em alternativa a redes de interconexão como Myrinet e lnfiniband. Experimentos com o modelo numérico de previsão do tempo WRF (Weather Research and Forecasting Model) e o algoritmo de granularidade fina IS do NAS Parallel Benchmarks, revelaram redução de mais de 20% no tempo de execução. Portanto, os resultados empíricos indicam um ganho expressivo no desempenho de um mesmo cluster quando configurado segundo a abordagem proposta, provando a pertinência deste trabalho.
Referências
AMO. Amd opteron(tm) processar product data sheet. Technical report, Publication 23932, 2007.
B. Amstrong, H. Bae, R. Eigenmann, F. Saied, M. Sayeed, and Y. Zheng. Hpc benchmarking and performance evaluation with realistiS applications. SPEC Benchmark Workslrop, 2006.
D. H. Bailey, H. Barszcz, J. T. Barton, D. S. Browning, R. L. Carter, L. Dagum, R. A. Fatoohi, P. O. Frederickson, T. A. Lasinski, R. S. Schreiber, H. D. Simmon, V. Venkatakrishnan, and S. K. Weeratunga. The nas parallel benchmarks. Intemational Joumal of Supercomputer Applications, 5(3):63-73, 1991.
N. Boden, D. Cohen, R. Felderman, A. Kulawik, C. Seitz, J. Seizovic, and W. Su. Myrinet: A gigabit-per-second local area network. IEEE Micro, 15(1):29-36, 1995.
R. Brightwell and K. Underwood. An analysis of the impact of mpi overlap and independent progress. International Conference on Supercomputing, 2004.
F. Cappello and D. Etiemble. Mpi versus mpi+Openmp on the ibm sp for the nas benchmarks. Supercomputing, 2000.
D. Cassiday. lnfiniband architecture. Hot Cllips 12, 2000.
L. Chai, A. Hartono, and D. Panda. Designing high performance and scalable mpi intra-node communication support for clusters. IEEE lnternational Conference on Cluster Computing, 2006.
G. Coulouris, J. Dollimore, and T. Kindberg. Distributed systems: Concepts and Design. Addison Wesley, 4" edition, 2005.
D. Dunning, G. Regnier, G. McAlpine, D. Cameron, B. Shubert, F. Berry, A. M. Merritt, E. Gronke, and C. Dodd. The virtual interface architecture. IEEE Micro, 18(2):66-76, 1998.
A. Faraj and X. Yuan. Communication characteristics in the nas parallel benchmarks. Parallel and Distributed Computing and Systems, 2002.
J. L. Hennessy and D. A. Panerson. Computer Architecture - A Quantitative Approach. Morgan Kaufmann Publishers, 3' edition, 2003.
Intel@. Intel® xeon® processar with 533 mhz fsb at2ghz to 3.20ghz datasheet. Technical report, Publ. 252135, 2004.
O. Kerbyson, K. Barker, and K. Davis. Analys is of the weather research and forcasting (wrf) model on large-scale systems. 2007.
J. Kim and D. Lilja. Characterization o f communication patterns in message-passing parallel scientific application programs. Communication, Architecture, and Applications for Network-Based Parallel Computing, pages 202-216, 1998.
V. Kumar, A. Grama, A. Gupta, and G. Karypis. Introduction to Parallel Computing. The Benjamin/Cummings Publishing Company Inc., 1º cdition, 1994.
M. Lobosco, V. S. Costa, and C. L. de Amorim. Performance evaluation of fast ethemet, giganet and myrinet on a cluster. International Conference on Computational Science, pages 296-305, 2002.
P. Luszczek, D. Bailey, 1. Dongarra, J. Kepner, R. Lucas, R. Rabenseifner, and D. Takahashi. The hpc challenge (hpcc) benchmark suite. IEEE SC06 Conference Tutorial, 2006.
R. Martin. A Systematic Characterization of Application Sensitivity to Network Performance. PhD thesis, Berkeley, 1999.
H. Meuer, E. Strohmaier, J. Dongarra, H. D. Simon, U. of Mannheim, and U. of Tennessee. Top500 supercomputing sites (http://www.top500.org), 2008.
J. Michalakes, J. Dudhia, D. Gill, T. Henderson, J. Klemp, W. Skamarock, and W. Wang. The weather research and forecast model: Software architecture and performance. ECMWF Workshop on the Use of High Perfonnance Computing in Meteorology, 2004.
J. Michalakes, J. Dudhia, D. Gill, J. Klemp, and W. Skamarock. Design of a next-generation regional weather research and forecast model. Towards Teracomputing, World Scientific, pages 117-124, 1999.
L. C. Pinto, R. P. Mendonça, and M. A. R. Dantas. The impact of interconnection networks and applications granularity to compound cluster configurations. IEEE Symposium on Computers and Communications, 2008.
H. Pourreza and P. Graham. On the programming impact of multi-core, multi-processar nodes in mpi clusters. High Performance Computing Systems and Applications, 2007.
R. Rabenseifner and A. E. Koniges. The parallel communication and i/o bandwidth benchmarks: b_eff and b_efL_io. Cray User Group Conference, CUG Summit, 2001.
R. Rabenseifner and G. Wellein. Communication and optimization aspects of parallel programming models on hybrid architectures. International Journal of High Performance Computing Applications, 17( I ):49-62, 2003.
J. Subhlok, S. Venkataramaiah, and A. Singh. Characterizing nas benchmark performance on shared heterogeneous networks. IEEE lnternational Parallel and Distributed Processing Symposium, pages 86-94, 2002.
Y. Sun, J. Wang, and Z. Xu. Architetural implications of the nas mg and ft parallel benchmarks. Advances in Parallel and Distributed Computing, pages 235-240, 1997.
T. Tabe and Q. Stout. The use of mpi communication library in the nas parallel benchmarks. Technical report. Technical Report CSE-TR-386-99, University of Michigan, 1999.
R. Zamani and A. Afsahi. Communication characteristics of message-passing scientific and engineering applications. International Conference on Parallel and Distributed Computing and Systems (PDCS), pages 644-649, 2005.
B. Amstrong, H. Bae, R. Eigenmann, F. Saied, M. Sayeed, and Y. Zheng. Hpc benchmarking and performance evaluation with realistiS applications. SPEC Benchmark Workslrop, 2006.
D. H. Bailey, H. Barszcz, J. T. Barton, D. S. Browning, R. L. Carter, L. Dagum, R. A. Fatoohi, P. O. Frederickson, T. A. Lasinski, R. S. Schreiber, H. D. Simmon, V. Venkatakrishnan, and S. K. Weeratunga. The nas parallel benchmarks. Intemational Joumal of Supercomputer Applications, 5(3):63-73, 1991.
N. Boden, D. Cohen, R. Felderman, A. Kulawik, C. Seitz, J. Seizovic, and W. Su. Myrinet: A gigabit-per-second local area network. IEEE Micro, 15(1):29-36, 1995.
R. Brightwell and K. Underwood. An analysis of the impact of mpi overlap and independent progress. International Conference on Supercomputing, 2004.
F. Cappello and D. Etiemble. Mpi versus mpi+Openmp on the ibm sp for the nas benchmarks. Supercomputing, 2000.
D. Cassiday. lnfiniband architecture. Hot Cllips 12, 2000.
L. Chai, A. Hartono, and D. Panda. Designing high performance and scalable mpi intra-node communication support for clusters. IEEE lnternational Conference on Cluster Computing, 2006.
G. Coulouris, J. Dollimore, and T. Kindberg. Distributed systems: Concepts and Design. Addison Wesley, 4" edition, 2005.
D. Dunning, G. Regnier, G. McAlpine, D. Cameron, B. Shubert, F. Berry, A. M. Merritt, E. Gronke, and C. Dodd. The virtual interface architecture. IEEE Micro, 18(2):66-76, 1998.
A. Faraj and X. Yuan. Communication characteristics in the nas parallel benchmarks. Parallel and Distributed Computing and Systems, 2002.
J. L. Hennessy and D. A. Panerson. Computer Architecture - A Quantitative Approach. Morgan Kaufmann Publishers, 3' edition, 2003.
Intel@. Intel® xeon® processar with 533 mhz fsb at2ghz to 3.20ghz datasheet. Technical report, Publ. 252135, 2004.
O. Kerbyson, K. Barker, and K. Davis. Analys is of the weather research and forcasting (wrf) model on large-scale systems. 2007.
J. Kim and D. Lilja. Characterization o f communication patterns in message-passing parallel scientific application programs. Communication, Architecture, and Applications for Network-Based Parallel Computing, pages 202-216, 1998.
V. Kumar, A. Grama, A. Gupta, and G. Karypis. Introduction to Parallel Computing. The Benjamin/Cummings Publishing Company Inc., 1º cdition, 1994.
M. Lobosco, V. S. Costa, and C. L. de Amorim. Performance evaluation of fast ethemet, giganet and myrinet on a cluster. International Conference on Computational Science, pages 296-305, 2002.
P. Luszczek, D. Bailey, 1. Dongarra, J. Kepner, R. Lucas, R. Rabenseifner, and D. Takahashi. The hpc challenge (hpcc) benchmark suite. IEEE SC06 Conference Tutorial, 2006.
R. Martin. A Systematic Characterization of Application Sensitivity to Network Performance. PhD thesis, Berkeley, 1999.
H. Meuer, E. Strohmaier, J. Dongarra, H. D. Simon, U. of Mannheim, and U. of Tennessee. Top500 supercomputing sites (http://www.top500.org), 2008.
J. Michalakes, J. Dudhia, D. Gill, T. Henderson, J. Klemp, W. Skamarock, and W. Wang. The weather research and forecast model: Software architecture and performance. ECMWF Workshop on the Use of High Perfonnance Computing in Meteorology, 2004.
J. Michalakes, J. Dudhia, D. Gill, J. Klemp, and W. Skamarock. Design of a next-generation regional weather research and forecast model. Towards Teracomputing, World Scientific, pages 117-124, 1999.
L. C. Pinto, R. P. Mendonça, and M. A. R. Dantas. The impact of interconnection networks and applications granularity to compound cluster configurations. IEEE Symposium on Computers and Communications, 2008.
H. Pourreza and P. Graham. On the programming impact of multi-core, multi-processar nodes in mpi clusters. High Performance Computing Systems and Applications, 2007.
R. Rabenseifner and A. E. Koniges. The parallel communication and i/o bandwidth benchmarks: b_eff and b_efL_io. Cray User Group Conference, CUG Summit, 2001.
R. Rabenseifner and G. Wellein. Communication and optimization aspects of parallel programming models on hybrid architectures. International Journal of High Performance Computing Applications, 17( I ):49-62, 2003.
J. Subhlok, S. Venkataramaiah, and A. Singh. Characterizing nas benchmark performance on shared heterogeneous networks. IEEE lnternational Parallel and Distributed Processing Symposium, pages 86-94, 2002.
Y. Sun, J. Wang, and Z. Xu. Architetural implications of the nas mg and ft parallel benchmarks. Advances in Parallel and Distributed Computing, pages 235-240, 1997.
T. Tabe and Q. Stout. The use of mpi communication library in the nas parallel benchmarks. Technical report. Technical Report CSE-TR-386-99, University of Michigan, 1999.
R. Zamani and A. Afsahi. Communication characteristics of message-passing scientific and engineering applications. International Conference on Parallel and Distributed Computing and Systems (PDCS), pages 644-649, 2005.
Publicado
29/10/2008
Como Citar
PINTO, Luiz C.; TOMAZELLA, Luiz H. B.; DANTAS, M. A. R..
Uma Abordagem para Composição de Clusters Eficientes na Execução do Modelo Numérico WRF de Previsão do Tempo. In: SIMPÓSIO EM SISTEMAS COMPUTACIONAIS DE ALTO DESEMPENHO (SSCAD), 9. , 2008, Campo Grande.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2008
.
p. 84-90.
DOI: https://doi.org/10.5753/wscad.2008.17671.
