E2SC 2017 Program
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Session 1 | Chair: Kevin J. Barker, Pacific Northwest National Lab Location: 506 |
| 9:00 AM - 9:10 AM | Welcome and Workshop Introduction |
| 9:10 AM - 10:00 AM | Keynote: Confessions of an Accidental Greenie: From Green Destiny to the Green500 and Beyond slides |
| Wu Feng, Virginia Tech | |
| 10:00 AM - 10:30 AM | Break |
| Session 2 | Chair: Joshua D. Suetterlein, Pacific Northwest National Lab Location: 506 |
| 10:30 AM - 11:00 AM | PoLiMEr: An Energy Monitoring and Power Limiting Interface for HPC Applications |
| Ivana Marincic, Venkatram Vishwanath, Henry Hoffmann | |
| University of Chicago, Argonne National Laboratory | |
| 11:00 AM - 11:30 AM | An Empirical Study of Performance and Energy Variation on Intel Processors |
| Amiruddha Marathe, Yijia Zhang, Grayson Blanks, Nirmal Kumbhare, Ghaleb Abdulla, Barry Rountree | |
| Lawrence Livermore National Lab, Boston University, University of Arizona | |
| 11:30 AM - 12:00 PM | Simulating Power Scheduling at Scale |
| Daniel Ellsworth, Tapasya Patki, Martin Schulz, Barry Rountree, Allen Malony | |
| Colorado College, Lawrence Livermore National Lab, University of Oregon | |
| 12:00 PM - 12:30 PM | Execution Phase Prediction Based on Phase Precursors and Locality |
| Saman Khoshbakht, Nikitas Dimopoulos | |
| University of Victoria | |
| 12:30 PM - 1:30 PM | Lunch Not Provided |
| Session 3 | Chair: Kevin J. Barker, Pacific Northwest National Lab Location: 506 |
| 1:30 PM - 2:00 PM | Scalable Performance Bounding Under Multiple Constrained Renewable Resources |
| Ramy Medhat, Shelby Funk, Barry Rountree | |
| University of Waterloo, University of Georgia, Lawrence Livermore National Lab | |
| 2:00 PM - 2:30 PM | Dynamic Application-aware Power Capping |
| Bo Wang, Dirk Schmidl, Christian Terboven, Malthias S. Muller | |
| RWTH Aachen University IT Center | |
| 2:30 PM - 3:00 PM | PANN: Power Allocation via Neural Networks: Dynamic Bounded-Power Allocation in High Performance Computing |
| Will Whiteside, Shelby Funk, Aniruddha Marathe, Barry Rountree | |
| University of Georgia, Lawrence Livermore National Lab | |
| 3:00 PM - 3:30 PM | Break |
| Session 4 | Chair: Ang Li, Pacific Northwest National Lab Location: 506 |
| 3:30 PM - 4:00 PM | Improving Energy Efficiency in Memory-constrained Applications using Core-specific Power Control |
| Sridutt Bhalachandra, Allan Poterfield, Stephen L. Olivier, Jan F. Prins, Robert J. Fowler | |
| UNC Chapel Hill, Sandia National Lab | |
| 4:00 PM - 4:30 PM | Adaptive Time-based Encoding for Energy-efficient Large Cache Architectures |
| Payman Behnam, Naser Sedaghati, Mahdi Nazm Bojnordi | |
| University of Utah, Imagination Tech | |
| 4:30 PM - 5:00 PM | Performance and Power Characteristics and Optimizations of Hybrid MPI/OpenMP LULESH Miniapps under Various Workloads |
| Xingfu Wu, Valerie Taylor, Jeanine Cook, Tanner Juedeman | |
| Argonne National Laboratory, Sandia National Laboratory, New Mexico State University | |
| Session 5 | Closing remarks Location: 506 |
Keynote InformationTitle: Confessions of an Accidental Greenie: From Green Destiny to the Green500 and Beyond Abstract While green computing appears to be mainstream, this was not always the case. It was only a decade ago when green computing, and in particular, green supercomputing, was still being scoffed at. Back then, supercomputing only focused on performance in terms of speed, as evidenced by the annual Gordon Bell Awards at Supercomputing (SC). Such a view is akin to purchasing an automobile based solely on its top speed rather than on other performance metrics such as fuel (energy) efficiency. Consequently, when Green Destiny debuted in April 2002 as (arguably) the world's first green supercomputer, it was resoundingly ridiculed --- so much so that one supercomputing visionary joked that "Green Destiny is so low power that it runs just as fast as when it is unplugged." At the same time, the audacity of Green Destiny, a 240-node supercomputer in 5.0 square feet and consuming as little as 3.2 kilowatts of power (or the equivalent of two hairdryers), created such a fervor as a disruptive technology that it led to international news coverage by the New York Times, CNN, the International Herald Tribune, PC World, Slashdot, and BBC News. Now 15 years later, this talk chronicles the confessions of an accidental greenie --- from how Green Destiny came about to how it evolved into the Green500 and what's next in green supercomputing. Bio Wu Feng is a Professor of Computer Science at Virginia Tech (VT), where he directs the Synergy Lab as well as the Synergistic Environments for Experimental Computing (SEEC) Center. He also holds appointments in the Department of Electrical and Computer Engineering, Health Sciences, Biomedical Engineering and Mechanics, and the Biocomplexity Institute. Dr. Feng has published 200+ manuscripts in high-performance computing, networking, storage, and analytics. Of recent note is the publication of The Green Computing Book: Tackling Energy Efficiency at Large Scale and a worldwide commercial on his research on biocomputing in the cloud, courtesy of the Microsoft Cloud. Dr. Feng holds a Ph.D. in Computer Science from the University of Illinois at Urbana-Champaign. In addition to being a Distinguished Scientist of the ACM, Dr. Feng has been named to HPCwire's Top People to Watch twice, once in 2004 and again in 2011. Presentation: slides | |
|
Proceedings available at the ACM Digital Library | |
