Computing on a dimeComputing in a warehouse 10 ² W10 W.

Panel:Energy Efficient Computing: hype or science?

Moderator: Feng Zhao, Microsoft ResearchPanelists:

Chuck Thacker, Microsoft ResearchFred Chong, UC Santa BarbaraRajesh Gupta, UC San DiegoPhilip Levis, Stanford UniversityTrishul Chilimbi, Microsoft Research

Computing and Energy

9 orders of magnitude in power difference

Tradeoffs in energy and performance across the scale

Computing on a dime Computing in a warehouse10 ⁻² W 10 ⁷ W

Issues to ponder

Systems vs componentsIf all the components are energy proportional, would the system automatically do the right thing?

Software vs hardwareWhat is the role of software in optimizing for energy efficiency?

Visibility and accountingDo we know where the joules go?

Energy vs performanceJust another proxy? Run as fast as possible?

Science vs engineeringDeeper roots of energy and power as related to computing?

Panelists

Chuck Thacker, MSRFred Chong, UC Santa BarbaraRajesh Gupta, UC San DiegoPhilip Levis, Stanford UniversityTrishul Chilimbi, MSR