Performance-Energy Trade-off in CMPs with Per-Core DVFS

  • Solomon Abera
  • M. Balakrishnan
  • Anshul Kumar
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10793)


In recent years, energy consumption of multicores has been a critical research agenda as chip multiprocessors (CMPs) have emerged as the leading architectural choice of computing systems. Unlike the uni-processor environment, the energy consumption of an application running on a CMP depends not only on the characteristics of the application but also the behavior of its co-runners (applications running on other cores). In this paper, we model the energy-performance trade-off using machine learning. We use the model to sacrifice a certain user-specified percentage of the maximum achievable performance of an application to save energy. The input to the model is the isolated memory behavior of the application and each of its co-runners, as well as the performance constraint. The output of the model is the minimum core frequency at which the application should run to guarantee the given performance constraint in the influence of the co-runners. We show that, in a quad-core processor, we can save up to 51% core energy by allowing 16% degradation of performance.


CMP Shared resource DVFS Machine learning 


  1. 1.
    Weiser, M., et al.: Scheduling for reduced CPU energy. USENIX (1994)Google Scholar
  2. 2.
    Zhu, D., Melhem, R., Childers, B.: Scheduling with dynamic voltage/speed adjustment using slack reclamation in multiprocessor real-time systems. IEEE TPDS 4, 686–700 (2003)Google Scholar
  3. 3.
    Cong, J., Gururaj, K.: Energy efficient multiprocessor task scheduling under input-dependent variation. In: DATE 2009, Dresden, Germany (2010)Google Scholar
  4. 4.
    Yao, F., et al.: A scheduling model for reduced CPU energy. In: FOCS 1995 (1995)Google Scholar
  5. 5.
    Ishihara, T., Yasuura, H.: Voltage scheduling problem for dynamically variable voltage processors. In: ISLPED 1998 (1998)Google Scholar
  6. 6.
    Kim, S.I., Kim, H.T., Kang, G.S., Kim, J.-K.: Using DVFS and task scheduling algorithms for a hard real-time heterogeneous multicore processor environment. In: EEHPDC 2013 (2013)Google Scholar
  7. 7.
    Zhuravlev, S., Blagodurov, S., Fedorova, A.: Addressing shared resource contention in multicore processors via scheduling. In: ASPLOS 2010 (2010)Google Scholar
  8. 8.
    Abera, S., Balakrishnan, M., Kumar, A.: PLSS: a scheduler for multi-core embedded systems. In: Knoop, J., Karl, W., Schulz, M., Inoue, K., Pionteck, T. (eds.) ARCS 2017. LNCS, vol. 10172, pp. 164–176. Springer, Cham (2017). CrossRefGoogle Scholar
  9. 9.
    Merkel, A., Stoess, J., Bellosa, F.: Resource-conscious scheduling for energy efficiency on multicore processors. In: EuroSys 2010 (2010)Google Scholar
  10. 10.
    Dhiman, G., Rosing, T.S.: Dynamic voltage frequency scaling for multi-tasking systems using online learning. In: ISLPED 2007 (2007)Google Scholar
  11. 11.
    Khan, U.A., Rinner, B.: Online learning of timeout policies for dynamic power management. ACM-TECS 13(4), 1–25 (2014)CrossRefGoogle Scholar
  12. 12.
    Otoom, M., et al.: Scalable and dynamic global power management for multicore chips. In: ACM 2015 (2015)Google Scholar
  13. 13.
    Ye, R., Xu, Q.: Learning-based power management for multicore processors via idle period manipulation. IEEE Trans. Comput.-Aided Des. Integr. Circuits Syst. 33, 1043–1055 (2014)CrossRefGoogle Scholar
  14. 14.
    Islam, F., Lin, M.: A framework for learning based DVFS technique selection and frequency scaling for multi-core real-time systems. In: HPCC 2015 (2015)Google Scholar
  15. 15.
    Shen, H., Qiu, Q.: Contention aware frequency scaling on CMPs with guaranteed quality of service. In: DATE 2014 (2014)Google Scholar
  16. 16.
    Hall, M., Frank, E., Holmes, G., Pfahringer, B., Reutemann, P., Witten, I.H.: The weka data mining software: an update. SIGKDD Explor. 11, 10–18 (2009)CrossRefGoogle Scholar
  17. 17.
    Breiman, L.: Random forest. Mach. Learn. 45(1), 5–32 (2001)CrossRefzbMATHGoogle Scholar
  18. 18.
    Li, S., et al.: McPAT: an integrated power, area, and timing modeling framework for multicore and manycore architectures. In: MICRO 2009 (2009)Google Scholar
  19. 19.
    Sniper Multicore Simulator.
  20. 20.
    Calder, B., et al.: SimPoint: picking representative samples to guide simulation (Chap. 7). In: Performance Evaluation and Benchmarking (2005)Google Scholar
  21. 21.
    Jaleel, A.: Memory characterization of workloads using instrumentation-driven simulation. Technical report, VSSAD (2007)Google Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Indian Institute of Technology DelhiNew DelhiIndia

Personalised recommendations