Dynamic power management using machine learning

ABSTRACT

Dynamic power management (DPM) work proposed to date places inactive components into low power states using a single DPM policy. In contrast, we instead dynamically select among a set of DPM policies with a machine learning algorithm. We leverage the fact that different policies outperform each other under different workloads and devices. Our algorithm adapts to changes in workloads and guarantees quick convergence to the best performing policy for each workload. We performed experiments with a policy set representing state of the art DPM policies on a hard disk drive and a WLAN card. Our results show that our algorithm adapts really well with changing device and workload characteristics and achieves an overall performance comparable to the best performing policy at any point of time.

REFERENCES

• 1: Luca Benini , Giovanni de Micheli, Dynamic Power Management: Design Techniques and CAD Tools, Kluwer Academic Publishers, Norwell, MA, 1998
• 2: Yoav Freund , Robert E. Schapire, A decision-theoretic generalization of on-line learning and an application to boosting, Journal of Computer and System Sciences, v.55 n.1, p.119-139, Aug. 1997 [doi>10.1006/jcss.1997.1504]
• 3: F. Douglis, P. Krishnan, and B. Bershad. Adaptive Disk Spin-Down Policies for Mobile Computers. In Computing Systems, volume 8, pages 381--413, 1995.
• 4: R. Golding, P. Bosch, and J. Wilkes. Idleness is not Sloth. In USENIX Winter Conference, pages 201--212, 1995.
• 5: Mani B. Srivastava , Anantha P. Chandrakasan , R. W. Brodersen, Predictive system shutdown and other architectural techniques for energy efficient programmable computation, IEEE Transactions on Very Large Scale Integration (VLSI) Systems, v.4 n.1, p.42-55, March 1996 [doi>10.1109/92.486080]
• 6: Chi-Hong Hwang , Allen C.-H. Wu, A predictive system shutdown method for energy saving of event-driven computation, Proceedings of the 1997 IEEE/ACM international conference on Computer-aided design, p.28-32, November 09-13, 1997, San Jose, California, United States
• 7: Eui-Young Chung , Luca Benini , Giovanni De Micheli, Dynamic power management using adaptive learning tree, Proceedings of the 1999 IEEE/ACM international conference on Computer-aided design, p.274-279, November 07-11, 1999, San Jose, California, United States
• 8: G. A. Paleologo , L. Benini , A. Bogliolo , G. De Micheli, Policy optimization for dynamic power management, Proceedings of the 35th annual conference on Design automation, p.182-187, June 15-19, 1998, San Francisco, California, United States [doi>10.1145/277044.277094]
• 9: Eui-Young Chung , Luca Benini , Alessandro Bogiolo , Giovanni De Micheli, Dynamic power management for non-stationary service requests, Proceedings of the conference on Design, automation and test in Europe, p.18-es, January 1999, Munich, Germany [doi>10.1145/307418.307456]
• 10: Qinru Qiu , Massoud Pedram, Dynamic power management based on continuous-time Markov decision processes, Proceedings of the 36th ACM/IEEE conference on Design automation, p.555-561, June 21-25, 1999, New Orleans, Louisiana, United States [doi>10.1145/309847.309997]
• 11: Tajana Simunic , Giovanni de Micheli , Luca Benini, Event-Driven Power Management of Portable Systems, Proceedings of the 12th international symposium on System synthesis, p.18, November 01-04, 1999
• 12: T. Simunic, L. Benini, and G. D. Micheli. Dynamic Power Management of Laptop Hard Disk. In Design Automation and Test in Europe, 2000.
• 13: Yung-Hsiang Lu , Eui-Young Chung , Tajana Šimunić , Luca Benini , Giovanni De Micheli, Quantitative comparison of power management algorithms, Proceedings of the conference on Design, automation and test in Europe, p.20-26, March 27-30, 2000, Paris, France [doi>10.1145/343647.343688]
• 14: A. Karlin, M. Manesse, L. McGeoch and S. Owicki. Competitive Randomized Algorithms for Nonuniform Problems. In Algorithmica, pp. 542--571, 1994.
• 15: C. Ruemmler and J. Wilkes. UNIX disk access patterns. In Proceedings of the Winter 1993 USENIX Conference, 1993.
• 16: Zhiyuan Ren , Bruce H. Krogh , Radu Marculescu, Hierarchical Adaptive Dynamic Power Management, Proceedings of the conference on Design, automation and test in Europe, p.10136, February 16-20, 2004
• 17: S. McFarling. Combining Branch Predictors. WRL Technical Note TN-36, Digital Equipment Corporation, June 1993.
• 18: Po-Ying Chang , Eric Hao , Yale N. Patt, Alternative implementations of hybrid branch predictors, Proceedings of the 28th annual international symposium on Microarchitecture, p.252-257, November 29-December 01, 1995, Ann Arbor, Michigan, United States

INDEX TERMS

Primary Classification:
  B. Hardware
  B.7 INTEGRATED CIRCUITS
      B.7.2 Design Aids

Additional Classification:
  B. Hardware
  B.8 Performance and Reliability
      B.8.2 Performance Analysis and Design Aids
  B.m MISCELLANEOUS
      Subjects: Design management

  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.6 Learning

General Terms:
Algorithms, Design, Performance

Keywords:
dynamic power management, machine learning