PampaFreq: Otimizando o EDP de Aplicações Paralelas em Processadores AMD
Resumo
O DVFS (Dynamic Voltage and Frequency Scaling) tem sido amplamente utilizado para melhorar o uso dos recursos computacionais quando aplicações paralelas estão sendo executadas. No entanto, as aplicações paralelas têm comportamentos distintos e se relacionam de diferentes maneiras com as políticas de modificação de frequência do DVFS. Neste sentido, é necessário utilizar métodos otimizados de DVFS para melhorar o custo-benefício entre desempenho e consumo de energia, representado pelo EDP (energy-delay product). Dito isso, através de uma extensa exploração de espaço e projeto de diferentes políticas de DVFS, níveis de frequência de operação da CPU e modo de operação de técnicas de boosting com a execução de dezesseis aplicações paralelas em três arquiteturas multicore, nós propomos PampaFreq, uma metodologia que otimiza o EDP em processadores AMD considerando as características da aplicação em tempo de execução. No caso mais significativo, PampaFreq otimiza o EDP em até 38% quando comparado com governor ondemand.Referências
Bailey, D. H., Barszcz, E., Barton, J. T., Browning, D. S., Carter, R. L., Dagum, L., Fatoohi, R. A., Frederickson, P. O., Lasinski, T. A., Schreiber, R. S., Simon, H. D., Venkatakrishnan, V., and Weeratunga, S. K. (1991). The nas parallel benchmarks & summary and preliminary results. In ACM/IEEE SC, pages 158–165, USA. ACM.
Basmadjian, R. and de Meer, H. (2018). Modelling and analysing conservative governor of dvfs-enabled processors. In Int. Conf. on Future Energy Systems, pages 519–525.
Branover, A., Foley, D., and Steinman, M. (2012). Amd fusion apu: Llano. Ieee Micro, 32(2):28–37.
Charles, J., Jassi, P., Ananth, N. S., Sadat, A., and Fedorova, A. (2009). Evaluation of the intel R(cid:13) coreTM i7 turbo boost feature. In IISWC, pages 188–197.
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 IEEE Int. Symp. on Workload Characterization, pages 44–54, DC, USA. IEEE Computer Society.
dos Santos Marques, W., de Souza, P. S. S., Lorenzon, A. F., Beck, A. C. S., Rutzig, M. B., and Rossi, F. D. (2017). Improving edp in multi-core embedded systems through multidimensional frequency scaling. In IEEE ISCAS, pages 1–4. IEEE.
Hackenberg, D., Ilsche, T., Schone, R., Molka, D., Schmidt, M., and Nagel, W. E. (2013). Power measurement techniques on standard compute nodes: A quantitative compari- son. In ISPASS, pages 194–204.
Hähnel, M., Döbel, B., Völp, M., and Härtig, H. (2012). Measuring energy consumption for short code paths using rapl. SIGMETRICS Perf. Evaluation Rev., 40(3):13–17.
Haj-Yahya, J., Mendelson, A., Ben Asher, Y., and Chattopadhyay, A. (2018). Power Management of Modern Processors, pages 1–55. Springer Singapore, Singapore.
Hsieh, C.-Y., Park, J.-G., Dutt, N., and Lim, S.-S. (2015). Memory-aware cooperative cpu-gpu dvfs governor for mobile games. In IEEE ESTIMedia, pages 1–8. IEEE.
Juan, D.-C., Garg, S., Park, J., and Marculescu, D. (2013). Learning the optimal operating point for many-core systems with extended range voltage/frequency scaling. In Int. Conf. on Hardware/Software Codesign and System Synthesis, page 8. IEEE Press.
Kondguli, S. and Huang, M. (2018). A case for a more effective, power-efcient turbo boosting. ACM Trans. on Architecture and Code Optimization, 15(1):1–22.
Lee, W. Y. (2009). Energy-saving dvfs scheduling of multiple periodic real-time tasks on multi-core processors. In Int. Symp. on Distributed Simulation and Real Time Appli- cations, pages 216–223.
Lo, D. and Kozyrakis, C. (2014). Dynamic management of turbomode in modern multi- core chips. In Int. Symp. on High Performance Computer Architecture, pages 603–613. IEEE.
Lorenzon, A. F. and Beck, A. C. S. (2019). Parallel Computing Hits the Power Wall - Principles, Challenges, and a Survey of Solutions. Springer Briefs in Computer Science. Springer.
Lorenzon, A. F., Dellagostin Souza, J., and Schneider Beck, A. C. (2017). Laant: A library to automatically optimize edp for openmp applications. In Design, Automation Test in Europe Conference Exhibition (DATE), 2017, pages 1229–1232.
Lorenzon, A. F., Sartor, A. L., Cera, M. C., and Schneider Beck, A. C. (2015). Optimized use of parallel programming interfaces in multithreaded embedded architectures. In 2015 IEEE Computer Society Annual Symposium on VLSI, pages 410–415.
Maiti, S., Kapadia, N., and Pasricha, S. (2015). Process variation aware dynamic power management in multicore systems with extended range voltage/frequency scaling. In IEEE Int. Midwest Symposium on Circuits and Systems (MWSCAS), pages 1–4. IEEE.
Marques, S. M. V. N., Medeiros, T. S., Rossi, F. D., Luizelli, M. C., Girardi, A. G., Beck, A. C. S., and Lorenzon, A. F. (2019). The impact of turbo frequency on the energy, performance, and aging of parallel applications. In 27th IFIP/IEEE International Con- ference on Very Large Scale Integration, VLSI-SoC 2019, Cuzco, Peru, October 6-9, 2019, pages 149–154. IEEE.
Raghavan, A., Luo, Y., Chandawalla, A., Papaefthymiou, M., Pipe, K. P., Wenisch, T. F., and Martin, M. M. (2012). Computational sprinting. In IEEE international symposium on high-performance comp architecture, pages 1–12. IEEE.
Sartor, A. L., Lorenzon, A. F., Carro, L., Kastensmidt, F., Wong, S., and Beck, A. C. S. (2017). Exploiting idle hardware to provide low overhead fault tolerance for vliw processors. J. Emerg. Technol. Comput. Syst., 13(2).
Wamhoff, J.-T., Diestelhorst, S., and Fetzer, C. The turbo diaries: Application-controlled frequency scaling explained.
Basmadjian, R. and de Meer, H. (2018). Modelling and analysing conservative governor of dvfs-enabled processors. In Int. Conf. on Future Energy Systems, pages 519–525.
Branover, A., Foley, D., and Steinman, M. (2012). Amd fusion apu: Llano. Ieee Micro, 32(2):28–37.
Charles, J., Jassi, P., Ananth, N. S., Sadat, A., and Fedorova, A. (2009). Evaluation of the intel R(cid:13) coreTM i7 turbo boost feature. In IISWC, pages 188–197.
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 IEEE Int. Symp. on Workload Characterization, pages 44–54, DC, USA. IEEE Computer Society.
dos Santos Marques, W., de Souza, P. S. S., Lorenzon, A. F., Beck, A. C. S., Rutzig, M. B., and Rossi, F. D. (2017). Improving edp in multi-core embedded systems through multidimensional frequency scaling. In IEEE ISCAS, pages 1–4. IEEE.
Hackenberg, D., Ilsche, T., Schone, R., Molka, D., Schmidt, M., and Nagel, W. E. (2013). Power measurement techniques on standard compute nodes: A quantitative compari- son. In ISPASS, pages 194–204.
Hähnel, M., Döbel, B., Völp, M., and Härtig, H. (2012). Measuring energy consumption for short code paths using rapl. SIGMETRICS Perf. Evaluation Rev., 40(3):13–17.
Haj-Yahya, J., Mendelson, A., Ben Asher, Y., and Chattopadhyay, A. (2018). Power Management of Modern Processors, pages 1–55. Springer Singapore, Singapore.
Hsieh, C.-Y., Park, J.-G., Dutt, N., and Lim, S.-S. (2015). Memory-aware cooperative cpu-gpu dvfs governor for mobile games. In IEEE ESTIMedia, pages 1–8. IEEE.
Juan, D.-C., Garg, S., Park, J., and Marculescu, D. (2013). Learning the optimal operating point for many-core systems with extended range voltage/frequency scaling. In Int. Conf. on Hardware/Software Codesign and System Synthesis, page 8. IEEE Press.
Kondguli, S. and Huang, M. (2018). A case for a more effective, power-efcient turbo boosting. ACM Trans. on Architecture and Code Optimization, 15(1):1–22.
Lee, W. Y. (2009). Energy-saving dvfs scheduling of multiple periodic real-time tasks on multi-core processors. In Int. Symp. on Distributed Simulation and Real Time Appli- cations, pages 216–223.
Lo, D. and Kozyrakis, C. (2014). Dynamic management of turbomode in modern multi- core chips. In Int. Symp. on High Performance Computer Architecture, pages 603–613. IEEE.
Lorenzon, A. F. and Beck, A. C. S. (2019). Parallel Computing Hits the Power Wall - Principles, Challenges, and a Survey of Solutions. Springer Briefs in Computer Science. Springer.
Lorenzon, A. F., Dellagostin Souza, J., and Schneider Beck, A. C. (2017). Laant: A library to automatically optimize edp for openmp applications. In Design, Automation Test in Europe Conference Exhibition (DATE), 2017, pages 1229–1232.
Lorenzon, A. F., Sartor, A. L., Cera, M. C., and Schneider Beck, A. C. (2015). Optimized use of parallel programming interfaces in multithreaded embedded architectures. In 2015 IEEE Computer Society Annual Symposium on VLSI, pages 410–415.
Maiti, S., Kapadia, N., and Pasricha, S. (2015). Process variation aware dynamic power management in multicore systems with extended range voltage/frequency scaling. In IEEE Int. Midwest Symposium on Circuits and Systems (MWSCAS), pages 1–4. IEEE.
Marques, S. M. V. N., Medeiros, T. S., Rossi, F. D., Luizelli, M. C., Girardi, A. G., Beck, A. C. S., and Lorenzon, A. F. (2019). The impact of turbo frequency on the energy, performance, and aging of parallel applications. In 27th IFIP/IEEE International Con- ference on Very Large Scale Integration, VLSI-SoC 2019, Cuzco, Peru, October 6-9, 2019, pages 149–154. IEEE.
Raghavan, A., Luo, Y., Chandawalla, A., Papaefthymiou, M., Pipe, K. P., Wenisch, T. F., and Martin, M. M. (2012). Computational sprinting. In IEEE international symposium on high-performance comp architecture, pages 1–12. IEEE.
Sartor, A. L., Lorenzon, A. F., Carro, L., Kastensmidt, F., Wong, S., and Beck, A. C. S. (2017). Exploiting idle hardware to provide low overhead fault tolerance for vliw processors. J. Emerg. Technol. Comput. Syst., 13(2).
Wamhoff, J.-T., Diestelhorst, S., and Fetzer, C. The turbo diaries: Application-controlled frequency scaling explained.
Publicado
21/10/2020
Como Citar
COSTA, Mariana; MARQUES, Sandro; MEDEIROS, Thiarles; ROSSI, Fábio; LUIZELLI, Marcelo; BECK, Antonio Carlos; LORENZON, Arthur.
PampaFreq: Otimizando o EDP de Aplicações Paralelas em Processadores AMD. In: SIMPÓSIO EM SISTEMAS COMPUTACIONAIS DE ALTO DESEMPENHO (SSCAD), 21. , 2020, Online.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2020
.
p. 49-60.
DOI: https://doi.org/10.5753/wscad.2020.14057.
