|
|
|
|
|
|
|
Microsoft explores efficiency
Environmental research efforts deepens
REDMOND, Wash. -- As part of its Sustainable Computing Program, Microsoft Corp. announced it will support four academic research projects focused on energy efficiency in computing in the areas of datacenter power efficiency, power management and the creation of parallel computing architecture with decreased power demands.
Through this program, Microsoft Research is stimulating research across a broad range of areas with the potential to significantly improve energy efficiency. And considering that a single 100-watt incandescent light bulb left on around the clock for a year costs more than $80 to power and releases 1,350 pounds of carbon dioxide emissions into the atmosphere -- roughly the same amount of power that an idle PC uses in that same timeframe -- finding ways to lessen humankind's impact on the environment is critical.
"We want to open new avenues of research and raise the awareness of power as a critical resource that needs to be managed," said Sailesh Chutani, senior director of Microsoft External Research. "Through this program, we are encouraging novel thinking about how to reduce that power consumption and how to make technology more environmentally friendly into the future."
The Sustainable Computing Program explores two main areas of research that can have a major impact. The first is the principle of "pay for play," which is the idea that the power consumed by a computing device should be proportional to the demand placed upon it, lowering the amount of energy consumed at low load and idle. Secondly, energy efficiency, even at peak loads, is equally important in reducing the overall consumption of electricity and should be managed as a first-class resource. The program encourages researchers to use novel approaches in hardware design, software, networking, benchmarking, analysis, virtualization and any other avenue that might provide improvements in the field.
Under the program, a total of $500,000 will be awarded among the four winners. A summary of the winners and descriptions of their projects follows:
* "Control-Theoretic Power and Performance Management for Green Data
Centers"; University of Tennessee; aimed at developing frameworks for
integrating power and performance improvements in virtualized
datacenters
* "Building a Building-scale Power Analysis Infrastructure"; Stanford
University; for the design and deployment of a dense sensor network for
power analysis, producing data for future research on power-aware
computing
* "A Synergistic Approach to Adaptive Power Management"; Harvard
University; for the development of a dynamic runtime environment that
ensures that power consumption is proportional to the computational
demands made on the system
* "Simulating Low Power x86 Architectures with Sooner, a Phoenix-based
Simulation Framework"; University of Oklahoma; for the development of a
simulation framework that supports the study of low-power
microarchitectures for innovative multicore systems
|
|
|
|
|
