• March 2012
    M T W T F S S

NSF’s Most Powerful Computing Resource Has Opened Its Doors to Six Science Teams

 National Science Foundation(03/21/12) Lisa-Joy Zgorski

 The U.S. National Science Foundation (NSF) and the University of Illinois’ National Center for Supercomputing Applications (NCSA) recently selected six teams to use the Blue Waters’ Early Science System before the full supercomputing system is deployed later this year.  “It began as an idea, and now thanks to sustained collaborative efforts by the entire project team, the vendor, and researchers, this computational tool is beginning to advance fundamental understanding in a wide range of scientific topics,” says NSF’s Irene Qualters.  NSF and NCSA awarded more than 24 research teams with time to use Blue Waters on compelling research queries.  A smaller group of six teams was chosen to use the Early Science System.  The six teams will pursue research in modeling high-temperature plasmas, simulating the formation and evolution of the Milky Way’s distant past, examining the protein that encases the HIV-1 genome, exploring explosive burning in Type 1a supernovae, and simulating the end of both the 20th and 21st centuries to explore changes in frequency and intensity of extreme events.  Once fully deployed, Blue Waters is expected to make arithmetic calculations at a sustained rate of more than 1 petaflop per second.


Scale-Out Processors: Bridging the Efficiency Gap Between Servers and Emerging Cloud Workloads

 HiPEAC (03/19/12)

Ecole Polytechnique Federale de Lausanne (EPFL) professor Babak Falsafi recently presented “Clearing the Clouds: A Study of Emerging Workloads on Modern Hardware,” which received the best paper award at ASPLOS 2012. “While we have been studying and tuning conventional server workloads (such as transaction processing and decision support) on hardware for over a decade, we really wanted to see how emerging scale-out workloads in modern data centers behave,” Falsafi says. “To our surprise, we found that much of a modern server processor’s hardware resources, including the cores, caches, and off-chip connectivity, are overprovisioned when running scale-out workloads leading to huge inefficiencies.” Efficiently executing scale-out workloads requires optimizing the instruction-fetch path for up to a few megabytes of program instructions, reducing the core complexity while increasing core counts, and shrinking the capacity of on-die caches to reduce area and power overheads, says EPFL Ph.D. student Mike Ferdman. The research was partially funded by the EuroCloud Server project. “Our goal is a 10-fold increase in overall server power efficiency through mobile processors and [three-dimensional] memory stacking,” says EuroCloud Server project coordinator Emre Ozer.