Energy-Efficient Computing

Tajana Simunic Rosing, Department of Computer Science and Engineering, University of California at San Diego, CA

Abstract: Energy consumption has become one of the most important design parameters in electronic systems. In high end systems the cost of energy powering the electronics equals the cost of cooling it. Battery powered and energy harvested systems have to ensure that there is enough power available to complete their tasks. In this talk I give an overview of the strategies whave developed at UCSD to significantly lower the energy consumption in both high and low end computing systems.

First I will discuss the tradeoffs between solely minimizing the power consumption versus combining that with thermal management in multicore systems. Next I will present how online learning can be used to adapt to changing workloads by selecting among a set of energy and thermal management policies. We achieve measured energy savings as high as 92% in a multitasking environment. In addition, there is a 20% and 60% average decrease in the frequency of hot spots and thermal cycles respectively in comparison to the best performing individual policy.

About the speaker: Tajana Simunic Rosing received her M.S. and PhD degrees from Stanford University in 2000 and 2001 respectively. She is currently an Assistant Professor at University of California, San Diego where she leads a research team. Her research interests include system-level hardware and software design for embedded, wireless and low-power systems.

More info