Ben Keller

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2015-257

December 18, 2015

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2015/EECS-2015-257.pdf

Dynamic voltage and frequency scaling (DVFS) can yield significant energy savings in processor SoCs. New integrated voltage regulator technologies enable fine-grained DVFS, in which many independent voltage domains switch voltage levels at nanosecond timescales to save energy. This work presents an overview of DVFS techniques and an energy model by which to evaluate them. Applying the energy model to traces of a cycle-accurate processor system simulation predicts energy savings of up to 53% from the application of fine-grained DVFS techniques. Further exploration and implementation of these technologies has the potential to dramatically reduce energy consumption in future systems.

Advisors: Borivoje Nikolic and Krste Asanović

---

BibTeX citation:

@mastersthesis{Keller:EECS-2015-257,
    Author= {Keller, Ben},
    Editor= {Nikolic, Borivoje and Asanović, Krste},
    Title= {Opportunities for Fine-Grained Adaptive Voltage Scaling to Improve System-Level Energy Efficiency},
    School= {EECS Department, University of California, Berkeley},
    Year= {2015},
    Month= {Dec},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2015/EECS-2015-257.html},
    Number= {UCB/EECS-2015-257},
    Abstract= {Dynamic voltage and frequency scaling (DVFS) can yield significant energy savings in processor SoCs. New integrated voltage regulator technologies enable fine-grained DVFS, in which many independent voltage domains switch voltage levels at nanosecond timescales to save energy. This work presents an overview of DVFS techniques and an energy model by which to evaluate them. Applying the energy model to traces of a cycle-accurate processor system simulation predicts energy savings of up to 53% from the application of fine-grained DVFS techniques. Further exploration and implementation of these technologies has the potential to dramatically reduce energy consumption in future systems.},
}

EndNote citation:

%0 Thesis
%A Keller, Ben 
%E Nikolic, Borivoje 
%E Asanović, Krste 
%T Opportunities for Fine-Grained Adaptive Voltage Scaling to Improve System-Level Energy Efficiency
%I EECS Department, University of California, Berkeley
%D 2015
%8 December 18
%@ UCB/EECS-2015-257
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2015/EECS-2015-257.html
%F Keller:EECS-2015-257