Mecanismo para reduzir o desperdício energético na pós-execução de aplicações em GPU
Resumo
Com o aumento da demanda por aceleradores como GPU para processamento de propósito geral em HPC, o impacto do consumo de energia destes recursos não pode ser negligenciado. Para reduzir o consumo de energia, algumas estratégias foram aplicadas, mas suas abordagens têm sido quase sempre focadas na economia de energia durante a execução da aplicação. Este trabalho é focado na economia de energia na pós-execução de aplicações. Quando o comportamento da pós-execução das aplicações é analisado em novas placas, observa-se que o consumo de energia não retorna rapidamente para o estado de repouso (idle) de maneira eficiente, gerando um desperdício de energia inesperado. De encontro a este processo ineficiente visando uma redução no desperdício de energia no período de pós-execução, foi desenvolvida uma estratégia para reduzir este desperdício com um impacto mínimo no desempenho global. Com essa estratégia, foi obtida uma economia de energia de até 15% nesta fase de pós-execução de uma aplicação executada sucessivas vezes. No caso de umaúnica execução, nossa abordagem permite uma poupança até 59% no consumo de energia na pós-execução.Referências
Agullo, E., Demmel, J., Dongarra, J., Hadri, B., Kurzak, J., Langou, J., Ltaief, H., Luszczek, P., and Tomov, S. (2009). Numerical linear algebra on emerging architectures: The plasma and magma projects. In Journal of Physics: Conference Series, volume 180, page 012037. IOP Publishing.
Che, S., Boyer, M., Meng, J., Tarjan, D., Sheaffer, J. W., Lee, S.-H., and Skadron, K. (2009). Rodinia: A benchmark suite for heterogeneous computing. In Workload Characterization, 2009. IISWC 2009. IEEE International Symposium on, pages 44–54. IEEE.
Che, S., Sheaffer, J. W., Boyer, M., Szafaryn, L. G., Wang, L., and Skadron, K. (2010). A characterization of the rodinia benchmark suite with comparison to contemporary cmp workloads. In Workload Characterization (IISWC), 2010 IEEE International Symposium on, pages 1–11. IEEE.
Chen, J., Li, B., Zhang, Y., Peng, L., and Peir, J.-k. (2011). Statistical gpu power analysis using tree-based methods. In Green Computing Conference and Workshops (IGCC), 2011 International, pages 1–6. IEEE.
Ge, R., Vogt, R., Majumder, J., Alam, A., Burtscher, M., and Zong, Z. (2013). Effects of dynamic voltage and frequency scaling on a k20 gpu. In 2nd International Workshop on Power-aware Algorithms, Systems, and Architectures.
Haidar, A., Tomov, S., Yamazaki, I., Solca, R., Schulthess, T., Dong, T., and Dongarra, J. (2008). Magma: A breakthrough in solvers for eigenvalue problems. http://icl. utk.edu/magma/.
Huang, S., Xiao, S., and Feng, W.-c. (2009). On the energy efciency of graphics processing units for scientic computing. In Parallel & Distributed Processing, 2009. IPDPS 2009. IEEE International Symposium on, pages 1–8. IEEE.
Kasichayanula, K., Terpstra, D., Luszczek, P., Tomov, S., Moore, S., and Peterson, G. D. (2012). Power aware computing on gpus. In Application Accelerators in High Performance Computing (SAAHPC), 2012 Symposium on, pages 64–73. IEEE.
Koomey, J. (2011). Growth in data center electricity use 2005 to 2010. Oakland, CA: Analytics Press. August, 1:2010.
Lee, J., Sathisha, V., Schulte, M., Compton, K., and Kim, N. S. (2011). Improving throughput of power-constrained gpus using dynamic voltage/frequency and core scaling. In Parallel Architectures and Compilation Techniques (PACT), 2011 International Conference on, pages 111–120. IEEE.
Ma, K., Li, X., Chen, W., Zhang, C., and Wang, X. (2012). Greengpu: A holistic approach to energy efciency in gpu-cpu heterogeneous architectures. In Parallel Processing (ICPP), 2012 41st International Conference on, pages 48–57. IEEE.
Mei, X., Yung, L. S., Zhao, K., and Chu, X. (2013). A measurement study of gpu dvfs on energy conservation. In Proceedings of the Workshop on Power-Aware Computing and Systems, page 10. ACM.
NVIDIA (2012). Whitepaper nvidia’s next generation cuda compute architecture: Kepler gk110. http://www.nvidia.com/content/PDF/kepler/ NVIDIA-Kepler-GK110-Architecture-Whitepaper.pdf.
NVIDIA (2013). Nvidia management library (nvml). https://developer. nvidia.com/nvidia-management-library-nvml.
Padoin, E. L., Pilla, L. L., Boito, F. Z., Kassick, R. V., Velho, P., and Navaux, P. O. (2012). Evaluating application performance and energy consumption on hybrid cpu+ gpu architecture. Cluster Computing, pages 1–15.
Tiwari, A., Laurenzano, M., Peraza, J., Carrington, L., and Snavely, A. (2012). Green queue: Customized large-scale clock frequency scaling. In Cloud and Green Computing (CGC) , 1-3 November , 2012, Xiangtan, Hunan, China. IEEE, IEEE.
Top500.org (2013). China’s tianhe-2 supercomputer maintains top spot on 42nd top500 list. http://www.top500.org/blog/lists/2013/11/press-release/.
Ukidave, Y. and Kaeli, D. R. (2013). Analyzing optimization techniques for power efIn Parallel and Distributed Processing Symciency on heterogeneous platforms. posium Workshops & PhD Forum (IPDPSW), 2013 IEEE 27th International, pages 1040–1049. IEEE.
Ukidave, Y., Ziabari, A., Mistry, P., Schirner, G., and Kaeli, D. (2013). Quantifying the energy efciency of fft on heterogeneous platforms. In Performance Analysis of Systems and Software (ISPASS), 2013 IEEE International Symposium on, pages 235– 244.
UTK, I. C. L. (2008). Matrix algebra on gpu and multicore architectures. http://icl.utk.edu/magma/.
Che, S., Boyer, M., Meng, J., Tarjan, D., Sheaffer, J. W., Lee, S.-H., and Skadron, K. (2009). Rodinia: A benchmark suite for heterogeneous computing. In Workload Characterization, 2009. IISWC 2009. IEEE International Symposium on, pages 44–54. IEEE.
Che, S., Sheaffer, J. W., Boyer, M., Szafaryn, L. G., Wang, L., and Skadron, K. (2010). A characterization of the rodinia benchmark suite with comparison to contemporary cmp workloads. In Workload Characterization (IISWC), 2010 IEEE International Symposium on, pages 1–11. IEEE.
Chen, J., Li, B., Zhang, Y., Peng, L., and Peir, J.-k. (2011). Statistical gpu power analysis using tree-based methods. In Green Computing Conference and Workshops (IGCC), 2011 International, pages 1–6. IEEE.
Ge, R., Vogt, R., Majumder, J., Alam, A., Burtscher, M., and Zong, Z. (2013). Effects of dynamic voltage and frequency scaling on a k20 gpu. In 2nd International Workshop on Power-aware Algorithms, Systems, and Architectures.
Haidar, A., Tomov, S., Yamazaki, I., Solca, R., Schulthess, T., Dong, T., and Dongarra, J. (2008). Magma: A breakthrough in solvers for eigenvalue problems. http://icl. utk.edu/magma/.
Huang, S., Xiao, S., and Feng, W.-c. (2009). On the energy efciency of graphics processing units for scientic computing. In Parallel & Distributed Processing, 2009. IPDPS 2009. IEEE International Symposium on, pages 1–8. IEEE.
Kasichayanula, K., Terpstra, D., Luszczek, P., Tomov, S., Moore, S., and Peterson, G. D. (2012). Power aware computing on gpus. In Application Accelerators in High Performance Computing (SAAHPC), 2012 Symposium on, pages 64–73. IEEE.
Koomey, J. (2011). Growth in data center electricity use 2005 to 2010. Oakland, CA: Analytics Press. August, 1:2010.
Lee, J., Sathisha, V., Schulte, M., Compton, K., and Kim, N. S. (2011). Improving throughput of power-constrained gpus using dynamic voltage/frequency and core scaling. In Parallel Architectures and Compilation Techniques (PACT), 2011 International Conference on, pages 111–120. IEEE.
Ma, K., Li, X., Chen, W., Zhang, C., and Wang, X. (2012). Greengpu: A holistic approach to energy efciency in gpu-cpu heterogeneous architectures. In Parallel Processing (ICPP), 2012 41st International Conference on, pages 48–57. IEEE.
Mei, X., Yung, L. S., Zhao, K., and Chu, X. (2013). A measurement study of gpu dvfs on energy conservation. In Proceedings of the Workshop on Power-Aware Computing and Systems, page 10. ACM.
NVIDIA (2012). Whitepaper nvidia’s next generation cuda compute architecture: Kepler gk110. http://www.nvidia.com/content/PDF/kepler/ NVIDIA-Kepler-GK110-Architecture-Whitepaper.pdf.
NVIDIA (2013). Nvidia management library (nvml). https://developer. nvidia.com/nvidia-management-library-nvml.
Padoin, E. L., Pilla, L. L., Boito, F. Z., Kassick, R. V., Velho, P., and Navaux, P. O. (2012). Evaluating application performance and energy consumption on hybrid cpu+ gpu architecture. Cluster Computing, pages 1–15.
Tiwari, A., Laurenzano, M., Peraza, J., Carrington, L., and Snavely, A. (2012). Green queue: Customized large-scale clock frequency scaling. In Cloud and Green Computing (CGC) , 1-3 November , 2012, Xiangtan, Hunan, China. IEEE, IEEE.
Top500.org (2013). China’s tianhe-2 supercomputer maintains top spot on 42nd top500 list. http://www.top500.org/blog/lists/2013/11/press-release/.
Ukidave, Y. and Kaeli, D. R. (2013). Analyzing optimization techniques for power efIn Parallel and Distributed Processing Symciency on heterogeneous platforms. posium Workshops & PhD Forum (IPDPSW), 2013 IEEE 27th International, pages 1040–1049. IEEE.
Ukidave, Y., Ziabari, A., Mistry, P., Schirner, G., and Kaeli, D. (2013). Quantifying the energy efciency of fft on heterogeneous platforms. In Performance Analysis of Systems and Software (ISPASS), 2013 IEEE International Symposium on, pages 235– 244.
UTK, I. C. L. (2008). Matrix algebra on gpu and multicore architectures. http://icl.utk.edu/magma/.
Publicado
08/10/2014
Como Citar
CARREÑO, Emmanuell; SARATES JR., Adiel; NAVAUX, Philippe.
Mecanismo para reduzir o desperdício energético na pós-execução de aplicações em GPU. In: SIMPÓSIO EM SISTEMAS COMPUTACIONAIS DE ALTO DESEMPENHO (SSCAD), 15. , 2014, São José dos Campos.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2014
.
p. 123-134.
DOI: https://doi.org/10.5753/wscad.2014.15005.
