publication . Conference object . Preprint . 2019

Q-Learning Inspired Self-Tuning for Energy Efficiency in HPC

Andreas Gocht; Mario Bielert; Robert Schöne;
Open Access
  • Published: 26 Jun 2019
  • Publisher: IEEE
Abstract
System self-tuning is a crucial task to lower the energy consumption of computers. Traditional approaches decrease the processor frequency in idle or synchronisation periods. However, in High-Performance Computing (HPC) this is not sufficient: if the executed code is load balanced, there are neither idle nor synchronisation phases that can be exploited. Therefore, alternative self-tuning approaches are needed, which allow exploiting different compute characteristics of HPC programs. The novel notion of application regions based on function call stacks, introduced in the Horizon 2020 Project READEX, allows us to define such a self-tuning approach. In this paper, ...
Subjects
free text keywords: Computer Science - Distributed, Parallel, and Cluster Computing, Computer Science - Machine Learning, Computer Science - Performance, Subroutine, Distributed computing, Idle, Supercomputer, Q-learning, Efficient energy use, Reinforcement learning, Energy consumption, Load balancing (computing), Computer science
Related Organizations
Funded by
EC| READEX
Project
READEX
Runtime Exploitation of Application Dynamism for Energy-efficient eXascale computing
  • Funder: European Commission (EC)
  • Project Code: 671657
  • Funding stream: H2020 | RIA
Communities
FET H2020FET HPC: HPC Core Technologies, Programming Environments and Algorithms for Extreme Parallelism and Extreme Data Applications
FET H2020FET HPC: Runtime Exploitation of Application Dynamism for Energy-efficient eXascale computing

[1] B. Rountree, D. K. Lownenthal, B. R. de Supinski, M. Schulz, V. W. Freeh, and T. Bletsch, “Adagio: Making DVS Practical for Complex HPC Applications,” in Proceedings of the 23rd International Conference on Supercomputing, 2009, DOI: 10.1145/1542275.1542340.

[2] R. Mijakovic, M. Firbach, and M. Gerndt, “An architecture for flexible auto-tuning: The Periscope Tuning Framework 2.0,” in 2016 2nd International Conference on Green High Performance Computing (ICGHPC), 2016, DOI: 10.1109/icghpc.2016.7508066. [OpenAIRE]

[3] D. Hackenberg, R. Schöne, T. Ilsche, D. Molka, J. Schuchart, and R. Geyer, “An Energy Efficiency Feature Survey of the Intel Haswell Processor,” in Parallel and Distributed Processing Symposium Workshop (IPDPSW), 2015 IEEE International, 2015, DOI: 10.1109/IPDPSW.2015.70. [OpenAIRE]

[4] G. Dhiman and T. Šimunic´ Rosing, “System-Level Power Management Using Online Learning,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, vol. 28, 2009, DOI: 10.1109/TCAD.2009.2015740. [OpenAIRE]

[5] H. Shen, J. Lu, and Q. Qiu, “Learning based DVFS for simultaneous temperature, performance and energy management,” in Thirteenth International Symposium on Quality Electronic Design (ISQED), 2012, DOI: 10.1109/ISQED.2012.6187575.

[6] A. Knüpfer, C. Rössel, D. a. Mey, S. Biersdorff, K. Diethelm, D. Eschweiler, M. Geimer, M. Gerndt, D. Lorenz, A. Malony, W. E. Nagel, Y. Oleynik, P. Philippen, P. Saviankou, D. Schmidl, S. Shende, R. Tschüter, M. Wagner, B. Wesarg, and F. Wolf, “Score-P: A Joint Performance Measurement Run-Time Infrastructure for Periscope,Scalasca, TAU, and Vampir,” in Tools for High Performance Computing 2011, 2012, DOI: 10.1007/978-3-642-31476-6_7. [OpenAIRE]

[7] R. Schöne, “A Unified Infrastructure for Monitoring and Tuning the Energy Efficiency of HPC Applications,” Ph.D. dissertation, Technischen Universität Dresden, 2017. [Online]. Available: https: //d-nb.info/1144299985/34 [OpenAIRE]

[8] J. Schuchart, M. Gerndt, P. G. Kjeldsberg, M. Lysaght, D. Horák, L. Rˇíha, A. Gocht, M. Sourouri, M. Kumaraswamy, A. Chowdhury, M. Jahre, K. Diethelm, O. Bouizi, U. S. Mian, J. Kružík, R. Sojka, M. Beseda, V. Kannan, Z. Bendifallah, D. Hackenberg, and W. E. Nagel, “The READEX formalism for automatic tuning for energy efficiency,” Computing, 2017, DOI: 10.1007/s00607-016-0532-7. [OpenAIRE]

[9] L. Rˇíha, J. Zapleta, M. Beseda, O. Vysocký, R. Schöne, A. Gocht, V. Kannan, K. Diethelm, J. C. Meyer, P. G. Kjeldsberg, M. Gerndt, and U. Locans, “D5.3 evaluation of the READEX Tool Suite using the READEX test-suite,” IT4I-VSB, TUD, ICHEC-NUIG, GNS, NTNU, TUM, INTEL, Tech. Rep., 2018. [Online]. Available: https://www.readex.eu/wp-content/uploads/2018/11/D5.3.pdf

[10] R. S. Sutton and A. G. Barto, Reinforcement learning : an introduction. Cambridge, Mass. [u.a.] : MIT Press, 2010.

[11] O. Vysocký, M. Beseda, L. Riha, J. Zapletal, V. Nikl, M. Lysaght, and V. Kannan, “Evaluation of the HPC Applications Dynamic Behavior in Terms of Energy Consumption,” in Proceedings of the Fifth International Conference on Parallel, Distributed, Grid and Cloud Computing for Engineering, 2017, DOI: 10.4203/ccp.111.3. [OpenAIRE]

[12] D. Hackenberg, T. Ilsche, J. Schuchart, R. Schöne, W. E. Nagel, M. Simon, and Y. Georgiou, “HDEEM: High Definition Energy Efficiency Monitoring,” in Energy Efficient Supercomputing Workshop (E2SC), 2014, 2014, DOI: 10.1109/E2SC.2014.13. [OpenAIRE]

Abstract
System self-tuning is a crucial task to lower the energy consumption of computers. Traditional approaches decrease the processor frequency in idle or synchronisation periods. However, in High-Performance Computing (HPC) this is not sufficient: if the executed code is load balanced, there are neither idle nor synchronisation phases that can be exploited. Therefore, alternative self-tuning approaches are needed, which allow exploiting different compute characteristics of HPC programs. The novel notion of application regions based on function call stacks, introduced in the Horizon 2020 Project READEX, allows us to define such a self-tuning approach. In this paper, ...
Subjects
free text keywords: Computer Science - Distributed, Parallel, and Cluster Computing, Computer Science - Machine Learning, Computer Science - Performance, Subroutine, Distributed computing, Idle, Supercomputer, Q-learning, Efficient energy use, Reinforcement learning, Energy consumption, Load balancing (computing), Computer science
Related Organizations
Funded by
EC| READEX
Project
READEX
Runtime Exploitation of Application Dynamism for Energy-efficient eXascale computing
  • Funder: European Commission (EC)
  • Project Code: 671657
  • Funding stream: H2020 | RIA
Communities
FET H2020FET HPC: HPC Core Technologies, Programming Environments and Algorithms for Extreme Parallelism and Extreme Data Applications
FET H2020FET HPC: Runtime Exploitation of Application Dynamism for Energy-efficient eXascale computing

[1] B. Rountree, D. K. Lownenthal, B. R. de Supinski, M. Schulz, V. W. Freeh, and T. Bletsch, “Adagio: Making DVS Practical for Complex HPC Applications,” in Proceedings of the 23rd International Conference on Supercomputing, 2009, DOI: 10.1145/1542275.1542340.

[2] R. Mijakovic, M. Firbach, and M. Gerndt, “An architecture for flexible auto-tuning: The Periscope Tuning Framework 2.0,” in 2016 2nd International Conference on Green High Performance Computing (ICGHPC), 2016, DOI: 10.1109/icghpc.2016.7508066. [OpenAIRE]

[3] D. Hackenberg, R. Schöne, T. Ilsche, D. Molka, J. Schuchart, and R. Geyer, “An Energy Efficiency Feature Survey of the Intel Haswell Processor,” in Parallel and Distributed Processing Symposium Workshop (IPDPSW), 2015 IEEE International, 2015, DOI: 10.1109/IPDPSW.2015.70. [OpenAIRE]

[4] G. Dhiman and T. Šimunic´ Rosing, “System-Level Power Management Using Online Learning,” IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, vol. 28, 2009, DOI: 10.1109/TCAD.2009.2015740. [OpenAIRE]

[5] H. Shen, J. Lu, and Q. Qiu, “Learning based DVFS for simultaneous temperature, performance and energy management,” in Thirteenth International Symposium on Quality Electronic Design (ISQED), 2012, DOI: 10.1109/ISQED.2012.6187575.

[6] A. Knüpfer, C. Rössel, D. a. Mey, S. Biersdorff, K. Diethelm, D. Eschweiler, M. Geimer, M. Gerndt, D. Lorenz, A. Malony, W. E. Nagel, Y. Oleynik, P. Philippen, P. Saviankou, D. Schmidl, S. Shende, R. Tschüter, M. Wagner, B. Wesarg, and F. Wolf, “Score-P: A Joint Performance Measurement Run-Time Infrastructure for Periscope,Scalasca, TAU, and Vampir,” in Tools for High Performance Computing 2011, 2012, DOI: 10.1007/978-3-642-31476-6_7. [OpenAIRE]

[7] R. Schöne, “A Unified Infrastructure for Monitoring and Tuning the Energy Efficiency of HPC Applications,” Ph.D. dissertation, Technischen Universität Dresden, 2017. [Online]. Available: https: //d-nb.info/1144299985/34 [OpenAIRE]

[8] J. Schuchart, M. Gerndt, P. G. Kjeldsberg, M. Lysaght, D. Horák, L. Rˇíha, A. Gocht, M. Sourouri, M. Kumaraswamy, A. Chowdhury, M. Jahre, K. Diethelm, O. Bouizi, U. S. Mian, J. Kružík, R. Sojka, M. Beseda, V. Kannan, Z. Bendifallah, D. Hackenberg, and W. E. Nagel, “The READEX formalism for automatic tuning for energy efficiency,” Computing, 2017, DOI: 10.1007/s00607-016-0532-7. [OpenAIRE]

[9] L. Rˇíha, J. Zapleta, M. Beseda, O. Vysocký, R. Schöne, A. Gocht, V. Kannan, K. Diethelm, J. C. Meyer, P. G. Kjeldsberg, M. Gerndt, and U. Locans, “D5.3 evaluation of the READEX Tool Suite using the READEX test-suite,” IT4I-VSB, TUD, ICHEC-NUIG, GNS, NTNU, TUM, INTEL, Tech. Rep., 2018. [Online]. Available: https://www.readex.eu/wp-content/uploads/2018/11/D5.3.pdf

[10] R. S. Sutton and A. G. Barto, Reinforcement learning : an introduction. Cambridge, Mass. [u.a.] : MIT Press, 2010.

[11] O. Vysocký, M. Beseda, L. Riha, J. Zapletal, V. Nikl, M. Lysaght, and V. Kannan, “Evaluation of the HPC Applications Dynamic Behavior in Terms of Energy Consumption,” in Proceedings of the Fifth International Conference on Parallel, Distributed, Grid and Cloud Computing for Engineering, 2017, DOI: 10.4203/ccp.111.3. [OpenAIRE]

[12] D. Hackenberg, T. Ilsche, J. Schuchart, R. Schöne, W. E. Nagel, M. Simon, and Y. Georgiou, “HDEEM: High Definition Energy Efficiency Monitoring,” in Energy Efficient Supercomputing Workshop (E2SC), 2014, 2014, DOI: 10.1109/E2SC.2014.13. [OpenAIRE]

Any information missing or wrong?Report an Issue