Abstract
This paper presents a Dynamic Power Management (DPM) framework based on reinforcement learning (RL) technique which aims to save power in an Energy-Managed Computer (EMC) system with self power-managed components. The proposed online adaptive DPM technique consists of two layers: component-level and system–level global power manager (GPM). The component-level PM policy is pre-specified and fixed whereas the system-level global PM employs temporal difference learning on Semi-Markov Decision Process (SMDP) for model-free RL, and it is specifically optimized for a multi-type application framework. Experiments show that that the proposed HPM scheme enhances power savings considerably while maintaining a good performance level. In comparison with other reference systems, the proposed RL DPM approach performs well under various workloads, can simultaneously consider power and performance and achieves a wide and deep power-performance tradeoff curves.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Benini, L., Bogliolo, A., De Micheli, G.: A survey of design techniques for system level dynamic power management. IEEE Trans. on VLSI Systems 8(3), 299–316 (2000)
Srivastava, M., Chandrakasan, A., Brodersen, R.: Predictive system shutdown and other architectural techniques for energy efficient programmable computation. IEEE Trans. on VLSI (1996)
Hwang, C.H., Wu, A.C.: A predictive system shutdown method for energy saving of event-driven computation. In: ICCAD 1997 (1997)
Benini, L., Paleologo, G., Bogliolo, A., De Micheli, G.: Policy optimization for dynamic power management. IEEE Trans. on CAD 18, 813–833 (1999)
Qiu, Q., Pedram, M.: Dynamic Power Management Based on Continuous-Time Markov Decision Processes. In: DAC 1999 (1999)
Simunic, T., Benini, L., Glynn, P., De Micheli, G.: Event-driven power management. IEEE Trans. on CAD (2001)
Jung, H., Pedram, M.: Dynamic power management under uncertain information. In: DATE 2007, pp. 1060–1065 (April 2007)
Dhiman, G., Simunic Rosing, T.: Dynamic power management using machine learning. In: ICCAD 2006, pp. 747–754 (November 2006)
Tan, Y., Liu, W., Qiu, Q.: Adaptive Power Management Using Reinforcement Learning. In: ICCAD 2009, pp. 461–467 (November 2009)
Wang, Y., Xie, Q., Ammari, A.C., Pedram, M.: Deriving a near-optimal power management policy using model-free reinforcement learning and Bayesian classification. In: DAC 2011, pp. 875–878 (June 2011)
Bradtke, S., Duff, M.: Reinforcement learning methods for continuous-time Markov decision problems. In: Advances in Neural Information Processing Systems 7, pp. 393–400. MIT Press (1995)
Bishop, C.M.: Pattern Recognition and Machine Learning. Springer (August 2006)
Lu, Y.-H., Benini, L., De Micheli, G.: Power-aware operating systems for interactive systems. IEEE Trans. VLSI System 10(4), 119–134 (2002)
Simunic, T., Boyd, S.: Managing power consumption in networks on chips. In: DATE 2002 (2002)
Sutton, R.S., Barto, A.G.: Reinforcement Learning: An Introduction. MIT Press, Cambridge (1998)
Rong, P., Pedram, M.: A Stochastic Framework for Hierarchical System-Level Power Management. In: Proc. of Symp. on Low Power Electronics and Design, pp. 269–274 (August 2005)
Ren, Z., Krogh, B.H., Marculescu, R.: Hierarchical adaptive dynamic power management. IEEE Trans. Computers 54(4), 409–420 (2005)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2013 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Triki, M., Wang, Y., Ammari, A.C., Pedram, M. (2013). Dynamic Power Management of a Computer with Self Power-Managed Components. In: Ayala, J.L., Shang, D., Yakovlev, A. (eds) Integrated Circuit and System Design. Power and Timing Modeling, Optimization and Simulation. PATMOS 2012. Lecture Notes in Computer Science, vol 7606. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-36157-9_22
Download citation
DOI: https://doi.org/10.1007/978-3-642-36157-9_22
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-36156-2
Online ISBN: 978-3-642-36157-9
eBook Packages: Computer ScienceComputer Science (R0)