Computação em Fluxos de Dados para Ambientes Paralelos e Heterogêneos
Resumo
As recentes mudanças no projeto de computadores tranformaram as plataformas de computação de alto desempenho em ambientes complexos equipados com diversos processadores paralelos e heterogêneos – como CPUs e GPUs. A utilização eficaz desses recursos continua sendo um grande desafio, pois essas plataformas adicionam novos compromissos e detalhes no projeto de aplicações. Nesse trabalho, apresentamos um conjunto de técnicas para maximizar a utilização de ambientes heterogêneos, que são desenvolvidas em nível de sistemas de execução e facilitam a exploração eficiente desses ambientes. Os resultados apresentam ganhos de aproximadamente 2× sobre as técnicas de escalonamento existentes na maioria dos experimentos.
Referências
Catalyurek, U., Ferreira, R., , Sachetto, R., and Teodoro, G. (2010). Scientific Computing with Multicore and Accelerators, chapter Dataflow Frameworks for Emerging Heterogeneous Architectures and Their Applications to Biomedicine. Chapman & Hall / CRC Press.
Ferreira, R., Jr., W. M., Guedes, D., Drummond, L., Coutinho, B., Teodoro, G., Tavares, T., Araujo, R., and Ferreira, G. (2005). Anthill:a scalable run-time environment for data mining applications. In SBAC-PAD.
Luk, C.-K., Hong, S., and Kim, H. (2009). Qilin: Exploiting parallelism on heterogeneous multiprocessors with adaptive mapping. In 42nd Int. Symp. on Microarchitecture (MICRO).
Teodoro, G. (2006). Suporte a fluxos de trabalho de aplicacões intensivas em dados (Best Master Thesis). CTD-SBAC/WSCAD. 2006.
Teodoro, G., Fireman, D., Guedes, D., Jr., W. M., and Ferreira, R. (2008a). Achieving multi-level parallelism in the filter-labeled stream programming model. Inter. Conf. on Parallel Processing (ICPP), 0:287–294.
Teodoro, G., Hartley, T. D. R., Catalyurek, U., and Ferreira, R. (2011). Optimizing dataflow applications on heterogeneous environments — accepted for publication. Cluster Computing: The Journal of Networks, Software Tools and Applications.
Teodoro, G., Hartley, T. D. R., Catalyurek, U. V., and Ferreira, R. (2010). Run-time optimizations for replicated dataflows on heterogeneous environments. In Proc. of the 19th ACM Inter. Symposium on High Performance Distributed Computing (HPDC) Best student paper.
Teodoro, G., Sachetto, R., Fireman, D., Guedes, D., and Ferreira, R. (2009a). Exploiting computational resources in distributed heterogeneous platforms. In 21st Intel Symposium on Computer Architecture and High Performance Computing (SBAC-PAD).
Teodoro, G., Sachetto, R., Sertel, O., Gurcan, M., Jr., W. M., Catalyurek, U., and Ferreira, R. (2009b). Coordinating the use of GPU and CPU for improving performance of compute intensive applications. In IEEE Int. Conf. on Cluster Computing.
Teodoro, G., Tavares, T., Ferreira, R., Kurc, T., Meira, W., Guedes, D., Pan, T., and Saltz, J. (2008b). A run-time system for efficient execution of scientific workflows on distributed environments. International Journal of Parallel Programming, 36.
Ferreira, R., Jr., W. M., Guedes, D., Drummond, L., Coutinho, B., Teodoro, G., Tavares, T., Araujo, R., and Ferreira, G. (2005). Anthill:a scalable run-time environment for data mining applications. In SBAC-PAD.
Luk, C.-K., Hong, S., and Kim, H. (2009). Qilin: Exploiting parallelism on heterogeneous multiprocessors with adaptive mapping. In 42nd Int. Symp. on Microarchitecture (MICRO).
Teodoro, G. (2006). Suporte a fluxos de trabalho de aplicacões intensivas em dados (Best Master Thesis). CTD-SBAC/WSCAD. 2006.
Teodoro, G., Fireman, D., Guedes, D., Jr., W. M., and Ferreira, R. (2008a). Achieving multi-level parallelism in the filter-labeled stream programming model. Inter. Conf. on Parallel Processing (ICPP), 0:287–294.
Teodoro, G., Hartley, T. D. R., Catalyurek, U., and Ferreira, R. (2011). Optimizing dataflow applications on heterogeneous environments — accepted for publication. Cluster Computing: The Journal of Networks, Software Tools and Applications.
Teodoro, G., Hartley, T. D. R., Catalyurek, U. V., and Ferreira, R. (2010). Run-time optimizations for replicated dataflows on heterogeneous environments. In Proc. of the 19th ACM Inter. Symposium on High Performance Distributed Computing (HPDC) Best student paper.
Teodoro, G., Sachetto, R., Fireman, D., Guedes, D., and Ferreira, R. (2009a). Exploiting computational resources in distributed heterogeneous platforms. In 21st Intel Symposium on Computer Architecture and High Performance Computing (SBAC-PAD).
Teodoro, G., Sachetto, R., Sertel, O., Gurcan, M., Jr., W. M., Catalyurek, U., and Ferreira, R. (2009b). Coordinating the use of GPU and CPU for improving performance of compute intensive applications. In IEEE Int. Conf. on Cluster Computing.
Teodoro, G., Tavares, T., Ferreira, R., Kurc, T., Meira, W., Guedes, D., Pan, T., and Saltz, J. (2008b). A run-time system for efficient execution of scientific workflows on distributed environments. International Journal of Parallel Programming, 36.
Publicado
19/07/2011
Como Citar
TEODORO, George; FERREIRA, Renato.
Computação em Fluxos de Dados para Ambientes Paralelos e Heterogêneos. In: CONCURSO DE TESES E DISSERTAÇÕES (CTD), 24. , 2011, Natal/RN.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2011
.
p. 76-81.
ISSN 2763-8820.
