InTheLoop | 02.06.2012
February 6, 2012
Berkeley Lab Breaks Ground on the CRT Facility
Energy Secretary Steven Chu, along with DOE, University of California, City of Berkeley, and Berkeley Lab leaders, broke ground on the Lab’s Computational Research and Theory (CRT) facility on Feb. 1. The CRT will be at the forefront of high performance computing research. Read more.
Can Cloud Computing Meet Scientific Needs? Yes, No, and Maybe
After a two-year study of the feasibility of cloud computing systems for meeting the ever-increasing computational needs of scientists, Department of Energy researchers have issued a report stating that the cloud computing model is useful, but should not replace the centralized supercomputing centers operated by DOE national laboratories. Read more.
SciDAC Outreach Center Participates in “Materials for Energy Applications” Workshop
Last week (Jan. 30 to Feb. 1) Berkeley Lab hosted an invitation-only workshop on Materials for Energy Applications, which was jointly sponsored by all 17 DOE national laboratories. This three-day conference—the first of a planned series—was held to increase industry awareness of relevant research capabilities within the DOE national laboratory system, to deepen the national laboratories’ understanding of the technical challenges facing industry, and to identify and improve paths forward for collaboration.
David Skinner of NERSC, who also heads the SciDAC Outreach Center, represented the Center with a poster on “Software Opportunities: Industry, ISVs [independent software vendors] and SciDAC.” The poster pointed out the breadth of the SciDAC software portfolio and how it can be leveraged in the manufacturing sector and the private sector more generally, increasing return on investment and decreasing time to solution.
Through a catalog of software that the center maintains and a list of “worked examples” from the five years of the SciDAC2 program, the poster is part of a larger effort to diversify the stakeholders in scientific software. “The impact of our investments in HPC software is measured in their end-to-end impacts over the long haul; they are not just about proof of principle,” said Skinner. “Finding ways for research-funded software to bring increased efficiency and innovation to a broad set of stakeholders beyond the SciDAC research community benefits all involved.”
“The past decades have seen a vigorous effort in bringing parallel computing to extreme problem sizes and extreme performance,” Skinner continued. “The time is ripe for the expertise developed within DOE/ASCR to be applied to the many parties who are newly interested in parallelism, whether that is motivated by the transition to multicore or the need to study systems of larger size. Expanding interest in parallel computing provides a way for the research community to see their work improved and sustained by that larger customer base.”
The last keynote address at the workshop was given by Michael McQuade, Senior Vice President for Science and Technology at United Technologies. The SciDAC Outreach Center and UTC won an IDC Innovation Award in 2011 for their application tuning work optimizing fuel spray nozzles to improve fuel efficiency. That outreach engagement yielded a 3.6x application speedup and a 66% reduction in nozzle design cycles. McQuade mentioned that he looks forward to the output of two similar joint projects that are currently under way.
Make Mine a Double: Moore's Law and the Future of Mathematics
What do iPhones, Twitter, Netflix, cleaner cities, safer cars, state-of-the-art environmental management, and modern medical diagnostics have in common? They are all made possible by Moore’s Law. In this editorial, David Bailey of Berkeley Lab's Computational Research Division and University of Newcastle mathematics professor Jon Borwein discuss trends in Moore's Law and the future of mathematics. Read more.
This Week’s Computing Sciences Seminars
Cloud Seminar: Science in the Clouds and Beyond
Monday, February 6, 11:00 am–12:00 pm, 380 Soda Hall, UC Berkeley
Lavanya Ramakrishnan, Computational Research Division, Lawrence Berkeley National Laboratory
Cloud computing has served the needs of web applications for the last few years. Scientific applications are increasingly investigating cloud models and technologies to manage their computation and data. Magellan, a project funded through the Department of Energy’s (DOE) Advanced Scientific Computing Research (ASCR) program, evaluated the use of cloud computing for scientific applications. In this talk, I will detail our results and experiences to date with cloud environments and technologies including the outcomes of various benchmarking and performance studies, cost analysis, and evaluation of MapReduce and other emerging cloud technologies.
Recent Developments of Fast Algorithms for Kohn-Sham Density Functional Theory
Wednesday, February 8, 4:10–5:00 pm, 939 Evans Hall, UC Berkeley
Lin Lin, Computational Research Division, Lawrence Berkeley National Laboratory
Kohn-Sham density functional theory (KSDFT) is the most widely used electronic structure theory for condensed matter systems. The standard method for solving KSDFT requires solving N eigenvectors for an O(N) * O(N) Kohn-Sham Hamiltonian matrix, with N being the number of electrons in the system. This procedure is expensive and scales as O(N^3). We have developed a pole expansion plus selected inversion (PEpSI) method, in which the KSDFT is solved by evaluating selected elements of the inverse of a series of sparse symmetric matrices, and the overall algorithm scales as at most O(N^2) for all materials including metallic and insulating systems. Recently we generalized the new method to nonorthogonal basis sets, with the electron density, total energy, Helmholtz free energy and atomic force calculated simultaneously and accurately. Combined with atomic orbital basis functions, the new method can be applied to study the electronic structure of boron nitride nanotube and carbon nanotube with more than 10,000 atoms on a single processor.
Accelerating Complex Software Using Custom Computing Machines
Friday, February 10, 1:00–2:00 pm, 50F-1647
Vinay Sriram, Rowland Institute at Harvard University
The field of computer hardware is going through a revolution. That era of ever increasing microprocessor clock speed has effectively come to an end. The focus now is on the use parallel energy efficient processors to accelerate complex software. A range of parallelization techniques and commodity parallel processors are now readily available. Moreover, at least from an Australian perspective, the inevitable introduction of the carbon tax is driving businesses towards reducing data centre power footprint. Custom computing machines, using FPGAs as acceleration devices, provide an energy efficient solution for high speed computing. In this presentation, two case studies will be presented where massive speedup was achieved using mathematical transformations, novel parallelization techniques and commodity parallel processors. A design space exploration of the benefits of using various commodity parallel processors with respect to price, power and silicon area utilization will also be presented.
About Computing Sciences at Berkeley Lab
The Lawrence Berkeley National Laboratory (Berkeley Lab) Computing Sciences organization provides the computing and networking resources and expertise critical to advancing the Department of Energy's research missions: developing new energy sources, improving energy efficiency, developing new materials and increasing our understanding of ourselves, our world and our universe.
ESnet, the Energy Sciences Network, provides the high-bandwidth, reliable connections that link scientists at 40 DOE research sites to each other and to experimental facilities and supercomputing centers around the country. The National Energy Research Scientific Computing Center (NERSC) powers the discoveries of 6,000 scientists at national laboratories and universities, including those at Berkeley Lab's Computational Research Division (CRD). CRD conducts research and development in mathematical modeling and simulation, algorithm design, data storage, management and analysis, computer system architecture and high-performance software implementation. NERSC and ESnet are DOE Office of Science User Facilities.
Lawrence Berkeley National Laboratory addresses the world's most urgent scientific challenges by advancing sustainable energy, protecting human health, creating new materials, and revealing the origin and fate of the universe. Founded in 1931, Berkeley Lab's scientific expertise has been recognized with 13 Nobel prizes. The University of California manages Berkeley Lab for the DOE’s Office of Science.
DOE’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States, and is working to address some of the most pressing challenges of our time. For more information, please visit science.energy.gov.