Accelerating Climate and Weather Simulations
Dr Shujia Zhou (NASA Goddard Space Flight Center, USA )
COMPUTER SCIENCE SEMINARDATE: 2010-05-21
TIME: 15:30:00 - 16:30:00
LOCATION: Ian Ross Seminar Room
CONTACT: JavaScript must be enabled to display this email address.
ABSTRACT:
Computer microprocessor development has been shifting from a single powerful core to multi- and many-cores, as shrinking chips approach physical limits. Thus, the era of continuous improvement in computing power fueled by the exponentially increasing speed of a single processor is now history. Yet applications such as Earth and space science simulators call for ever-higher resolutions and more sophisticated treatment of physical processes, making models even more computationally intensive. Therefore, it is crucial for the computational science and technology community to explore various ways of taking advantage of tremendous computing power provided by multi- and many-core processors, while alleviating the impacts of the new programming paradigm arising from this shift.
In the last few years, there have emerged unconventional processors (hereafter referred to as accelerators), such as IBMas Cell Broadband Engine (hereafter referred to as Cell) and NVIDIAas Graphics Processing Unit (GPU). Intel and AMD are also developing competing Cell or GPU-like processors, in addition to their conventional multi- and many-core processors. Certain computationally intensive applications with moderate data communication have demonstrated significant performance improvements on both Cell and GPU. However, these emerging processors require new programming paradigms, which increase the porting costs and impede their effective utilization. Therefore, hybrid computing is a practical approach to speed up those compute kernels in the accelerators, while running the remaining difficult-to-port application on the conventional processors. We will discuss our research in porting climate and weather applications to Cell and GPU as well as exploring the methods in connecting those accelerators to the conventional computer clusters.
BIO:
EDUCATION
1994 PhD, Physics, Washington University in St Louis
1987 MS, Physics, Institute of Metal Research, Chinese Academy of Sciences
1984 BS, Physics, Lanzhou University
EMPLOYMENT * 9/2008-present, Research Associate Professor, University of Maryland, Baltimore County * 1/2000-present, Department Scientist, Northrop Grumman Corporation, work at NASA Goddard Space Flight Center * 7/1999-1/2000, Senior Software Engineer, MYTA Corporation, work at Lockheed Martin Air Traffic Management Division * 10/1996-7/1999, Technical Staff Member, Los Alamos National Laboratory * 11/1994-10/1996, Director's Postdoctoral Fellow, Los Alamos National Laboratory * 11/1993-11/1994, Postdoctoral Research Associate, Virginia Tech * 7/1989-8/1990, Guest researcher, National Institute of Standards and Technology
SELECTED PUBLICATIONS
1. S. Zhou, et al, aLarge-Scale Molecular Dynamics Simulations of Dislocation Intersection in Copper,a Science, 279 1525 (1998). 2. S. Zhou, aCoupling Climate Models with Earth System Modeling Framework and Common Component Architecture,a Concurrency and Computation: Practice and Experience, 18 203 (2006).
3. L.C. McInnes, B.A. Allan, R. Armstrong, S.J. Benson, D.E. Bernholdt, T.L. Dahlgren, L.F. Diachin, M. Krishnan, J.A. Kohl, J.W. Larson, S. Lefantzi, J. Nieplocha, B. Norris, S.G. Parker, J. Ray, and S. Zhou, aParallel PDE-Based Simulations Using the Common Component Architecture," an invited chapter in the book aNumerical Solution of Partial Differential Equations on Parallel Computers,a A. M. Bruaset, P. Bjorstad, and A. Tveito, editors, Published by Springer-Verlag 2006.
4. S. Zhou et al., aHigh-Speed Network and Grid Computing for High-End Computation: Application in Geodynamics Ensemble Simulations,a Concurrency and Computation: Practice and Experience, 19 573 (2007).
5. S. Zhou et al., aCross-Organization Interoperability Experiments of Weather and Climate Models with the Earth System Modeling Framework,a Concurrency and Computation: Practice and Experience, 19 583 (2007). 6. S. Zhou, A. Oloso, M. Damon, T. Clune, aApplication Con-trolled Parallel Asynchronous IO,a Supercomputing Conference, Tampa, FL, U.S.A. (November 14, 2006)
7. S. Zhou and C. Cruz, aData Registration, Match, and Model Component Coupling,a Proceedings of the 22nd Annual ACM SIGPLAN Conference on Object-Oriented Programming, Systems, Languages, and Applications, OOPSLA 2007, October 2125, 2007, Montreal, Quebec, Canada.
8. S. Zhou, B.H. Van Aartsen, and T.L. Clune, aA Lightweight Scalable I/O Utility for Optimizing High-End Computing Applications,a Proceedings of the 22nd IEEE Parallel and Distributed Processing Symposium, Miami, April 14-18, 2008
9. S. Zhou, D. Duffy, T. Clune, M. Suarez, S. Williams, and M. Halem, aThe Impact of IBM Cell Technology on the Programming Paradigm in the Context of Computer Systems for Climate and Weather Models,a Concurrency and Computation: Practice and Experience, 21 2176 (2009).
10. S. Zhou, aSoftware Support for Hybrid Computing,a eScience Workshop 2009, Oct 15-17, 2009, Pittsburgh, Pennsylvania
11. S. Zhou, C. Cruz, D. Duffy, R. Tucker, M. Purcell, Accelerating Climate and Weather Simulations Through Hybrid Computing, The 10th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, May 17-20, 2010, Melbourne, Victoria, Australia
12. S. Zhou, I. Stewart, A Memory Centric Kernel Framework for Accelerating Short-Range, Interactive Particle Simulation, The 10th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, May 17-20, 2010, Melbourne, Victoria, Australia
13. Q. He, S. Zhou, B. Kobler, D. Duffy, T. McGlynn, Case Study for Running HPC Applications in Public Clouds, The ACM International Symposium on High Performance Distributed Computing, June 20-25, 2010, Chicago, Illinois
