Power Capping of CPU-GPU Heterogeneous Systems using Power and Performance Models

Kazuki Tsuzuku, Toshio Endo

Abstract

Recent high performance computing (HPC) systems and supercomputers are built under strict power budgets and the limitation will be even severer. Thus power control is becoming more important, especially on the systems with accelerators such as GPUs, whose power consumption changes largely according to the characteristics of application programs. In this paper, we propose an efficient power capping technique for compute nodes with accelerators that supports dynamic voltage frequency scaling (DVFS). We adopt a hybrid approach that consists of a static method and a dynamic method. By using a static method based on our power and performance model, we obtain optimal frequencies of GPUs and CPUs for the given application. Additionally, while the application is running, we adjust GPU frequency dynamically based on real-time power consumption. Through the performance evaluation on a compute node with a NVIDIA GPU, we demonstrate that our hybrid method successfully control the power consumption under a given power constraint better than simple methods, without aggravating energy-to-solution.

References

  1. Burtscher, M., Zecena, I., and Zong, Z. (2014). Measuring gpu power with the k20 built-in sensor. In Proceedings of Workshop on General Purpose Processing Using GPUs, pages 28:28-28:;36. ACM.
  2. Endo, T., Nukada, A., and Matsuoka, S. (2014). Tsubamekfc : a modern liquid submersion cooling prototype towards exascale becoming the greenest supercomputer in the world. In ICPADS2014, editor, The 20th IEEE International Conference on Parallel and Distributed Systems.
  3. Gandhi, A., Harchol-Balter, M., Das, R., and Lefurgy, C. (2009). Optimal power allocation in server farms. In Proceedings of the Eleventh International Joint Conference on Measurement and Modeling of Computer Systems, pages 157-168. ACM.
  4. Jiao, Y., Lin, H., Balaji, P., and W., F. (2010). Power and performance characterization of computational kernels on the gpu. In Proceedings of the 2010 IEEE/ACM Int'L Conference on Green Computing and Communications & Int'L Conference on Cyber, Physical and Social Computing, pages 221-228. IEEE.
  5. Komoda, T., Hayashi, S., Nakada, T., Miwa, S., and Nakamura, H. (2013). Power capping of cpu-gpu heterogeneous systems through coordinating dvfs and task mapping. In Computer Design (ICCD), IEEE 31st International Conference, pages 349-356. IEEE.
  6. Laros, J., e. a. (2014). High performance computing - power aoolication programming interface specification version 1.0. In SANDIA REPORT, volume SAND2014- 17061.
  7. Lucas, R. e. a. (2014). Top ten exascale research challenges. In DOE ASCAC Subcommitte Report.
  8. Matsuoka, S. (2011). Tsubame 2.0 begins - the long road from tsubame1.0 to 2.0(part two). In e Science Journal, editor, Global Scientific Information and Computing Center, volume 3, pages 2-8. Tokyo Institute of Technology.
  9. 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, pages 10:1-10:5. ACM.
  10. Nagasaka, H., Maruyama, N., Nukada, A., Endo, T., and Matsuoka, S. (2010). Statistical power modeling of gpu kernels using performance counters. In Green Computing Conference, 2010 International, pages 115-122. IEEE.
  11. Patki, T., Lowenthal, D. K., Rountree, B., Schulz, M., and de Supinski, B. R. (2013). Exploring hardware overprovisioning in power-constrained, high performance computing. In Proceedings of the 27th International ACM Conference on International Conference on Supercomputing, ICS'13, pages 173-182. ACM.
  12. Schöne, R., Treibig, J., Dolz, M. F., Guillen, C., Navarrete, C., Knobloch, M., and Rountree, B. (2014). Tools and methods for measuring and tuning the energy efficiency of hpc systems. In Scientific Programming, pages 273-283. IOS Press.
  13. Song, S., Su, C., Rountree, B., and Cameron, K. (2013). A simplified and accurate model of power-performance efficiency on emergent gpu architectures. In Parallel & Distributed Processing (IPDPS), pages 673-686. IEEE.
Download


Paper Citation


in Harvard Style

Tsuzuku K. and Endo T. (2015). Power Capping of CPU-GPU Heterogeneous Systems using Power and Performance Models . In Proceedings of the 4th International Conference on Smart Cities and Green ICT Systems - Volume 1: SMARTGREENS, ISBN 978-989-758-105-2, pages 226-233. DOI: 10.5220/0005445102260233


in Bibtex Style

@conference{smartgreens15,
author={Kazuki Tsuzuku and Toshio Endo},
title={Power Capping of CPU-GPU Heterogeneous Systems using Power and Performance Models},
booktitle={Proceedings of the 4th International Conference on Smart Cities and Green ICT Systems - Volume 1: SMARTGREENS,},
year={2015},
pages={226-233},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0005445102260233},
isbn={978-989-758-105-2},
}


in EndNote Style

TY - CONF
JO - Proceedings of the 4th International Conference on Smart Cities and Green ICT Systems - Volume 1: SMARTGREENS,
TI - Power Capping of CPU-GPU Heterogeneous Systems using Power and Performance Models
SN - 978-989-758-105-2
AU - Tsuzuku K.
AU - Endo T.
PY - 2015
SP - 226
EP - 233
DO - 10.5220/0005445102260233