A-Z Index | Phone Book | Careers

InTheLoop | 09.26.2011

September 26, 2011

Unlocking the Secret of Cellulose Deconstruction

Running a molecular dynamics simulation with atomistic details is like directing a vast crowd shot in a movie. So says Berkeley chemical and biomolecular engineering professor Jhih-Wei Chu. His crowd shots consist of thousands of molecular actors and take months to run even on the world’s best supercomputers. Although the action occurs at a scale invisible to the human eye, the outcome could have a dramatic impact on biofuel production. Chu's simulation required around one million CPU hours at NERSC. Read more.

A Better Lithium-Ion Battery

Lithium-ion batteries are everywhere, in smart phones, laptops, and an array of other consumer electronics. Good as they are, they could be much better, especially when it comes to lowering the cost and extending the range of electric cars. To do that, batteries need to store a lot more energy. A team of Berkeley Lab scientists have designed a new kind of anode — a critical energy-storing component — capable of absorbing eight times the lithium of current designs. Using supercomputers at NERSC, the team found a tailored polymer that conducts electricity and binds closely to lithium-storing silicon particles. Read more.

Annual DOE Workshop on HPC Best Practices Starts Today

The DOE Workshop on HPC Best Practices: File Systems and Archives is being held today and tomorrow, Sept. 26-27, in San Francisco. NERSC Storage Systems Group Lead Jason Hick is chairing this year’s workshop. More than 60 participants have come from across the U.S. as well as Europe and Japan.

This workshop will address current best practices for the procurement, operation, and usability of file systems and archives, and will address whether system challenges can be met by evolving current practices. A report will present findings to DOE and other stakeholders.

Brown and Bell Co-Organize DOE Applied Math Workshop

Computational Research Division Director David Brown and Center for Computational Sciences and Engineering (CCSE) head John Bell, as well as Mihai Anitescu (Argonne National Lab), Michael Ferris (Univ. of Wisconsin), and Robert Moser (Univ. of Texas, Austin), organized the recent DOE Applied Math Workshop on “Mathematics for the Analysis, Simulation, and Optimization of Complex Systems” that was held Sept., 13–14, 2011.

The goal of this workshop was to identify new research areas in applied mathematics that will complement and enhance the existing DOE ASCR Applied Mathematics Program efforts that address the understanding of complex systems. The report that will result from this workshop will be a valuable resource for defining the potential impact of research in applied mathematics on problems important to the mission of DOE. Details of the workshop can be found here.

Chen and Bell Kick Off New Combustion Exascale Co-Design Center

DOE announced the award of the Combustion Exascale Co-Design Center (CECDC) at the July SciDAC Meeting in Denver. The CECDC is led by Jackie Chen of Sandia as Director and Combution Lead, and John Bell of CCSE as Deputy Director and Math Lead. A kickoff meeting for the CECDC was held Sept. 20–22 at Sandia in Livermore. Marc Day, Vince Beckner, and Mike Lijewski of CCSE are also part of the new CECDC.

Two Alvarez Fellows in Computational Science Announced

Anubhav Jain and Lin Lin have both been awarded the 2011 Luis W. Alvarez Postdoctoral Fellowship in Computational Science. Hosted by the NERSC and Computational Research Divisions, the fellowship allows recent graduates with a Ph.D. (or equivalent) to acquire further scientific training at one of the leading facilities for scientific computing and to develop professional maturity for independent research. Read more.

New Recognition Awards Program for Berkeley Lab Employees

Berkeley Lab is launching a new annual awards program to honor exceptional achievements by laboratory employees. Two new types of awards, the Director’s Awards for Exceptional Achievement and The Berkeley Lab Prize — Lifetime Achievement Award, have been established by Director Paul Alivisatos to recognize teams and individuals for significant accomplishments advancing the Lab’s mission and strategic goals. Nominations will begin next month. Read more.

ICiS Is Accepting Proposals for Summer 2012 Workshop

The Institute for Computing in Science (ICiS) at Argonne National Laboratory is an innovative approach to accelerating science and engineering discoveries through advances in computational science and large-scale computing. ICiS is now accepting proposals for its summer 2012 workshop here.

Break Software Updated to Improve Performance

Use of a break reminder software can help you remember to take ergonomic breaks. RSI Guard by Remedy Interactive includes a break timer, customizable reminders, and personalization settings so you can tailor the reminders to your situation.

However, since the changeover at Apple to Intel chips, RSI Guard has had to run in the Rosetta Emulator, which degrades performance on a Mac. Some Mac users found that the software slowed down their systems. This issue has been resolved with the latest version of RSI Guard software for Macs, with native Intel (no Rosetta). The IT Division has updated both the Mac and Windows versions of RSI Guard software on the LBNL Software Download page.

If your system will not allow installation of a break reminder software, consider these tips for taking breaks:

  • Eyes: Every 20 minutes look off at a distance >20 feet, then look at something close for a moment and repeat several times.
  • Hands and arms: Every 20 minutes do a few finger fans, then rest your hands on your lap with thumbs-up and twiddle your thumbs a bit.
  • Seated body: Get up and stretch every 30 minutes. Get up and walk for five minutes every hour.

For any ergonomic questions, call the LBNL Ergo Team at x6848 or email them at ergo@lbl.gov.

MSRI Presents “From Math to Market and Back”

Friends of the Mathematical Sciences Research Institute (MSRI) are invited to a wide-ranging and fascinating discussion with Jim Simons, a mathematician, financier, and longtime MSRI trustee, who will be appearing in the next presentation of MSRI’s "Conversations with Bob Osserman” series. This free special event, titled “From Math to Market and Back: A Conversation with James Simons, Founder of the Fabled Medallion Fund,” will take place on Monday, October 3, from 7:00–8:30 pm in the auditorium of the Berkeley City College at 2050 Center Street. Please note that seating will be limited.

This Week’s Computing Sciences Seminars

AMPLab: Making Sense at Scale with Algorithms, Machines and People
Monday, Sept. 26, 3:30-5:00 pm, 3108 Etcheverry Hall, UC Berkeley
Michael Franklin, UC Berkeley

The creation, analysis, and dissemination of data have become profoundly democratized. Social networks spanning 100s of millions of users enable instantaneous discussion, debate, and information sharing. Streams of tweets, blogs, photos, and videos identify breaking events faster and in more detail than ever before. Deep, on-line datasets enable analysis of previously unreachable information.

The key challenge is that the massive scale and diversity of this continuous flood of information breaks our existing technologies. State-of-the-art Machine Learning algorithms do not scale to massive data sets. Existing data analytics frameworks cope poorly with incomplete and dirty data and cannot process heterogeneous multi-format information. Current large-scale processing architectures struggle with diversity of programming models and job types and do not support the rapid marshalling and unmarshalling of resources to solve specific problems. All of these limitations lead to a Scalability Dilemma: beyond a point, our current systems tend to perform worse as they are given more data, more processing resources, and involve more people — exactly the opposite of what should happen.

To address these issues, a group of us from machine learning, systems, databases, and networking have initiated a five-year research effort at Berkeley called the AMPLab, where AMP stands for "Algorithms, Machines, and People". AMPLab envisions a world where massive data, cloud computing, communication and people resources can be continually, flexibly and dynamically be brought to bear on a range of hard problems by huge numbers of people connected to the cloud via mobile and other client devices of increasing power and sophistication. The lab is supported by 15 leading technology companies including founding sponsors Google and SAP.

In this talk, I will give an overview of the AMPLab motivation and research agenda and discuss several of our initial projects. One such project, CrowdDB, is developing infrastructure to support hybrid cloud/crowd query answering systems — leveraging the very different skills, and performance, reliability, and cost characteristics of large groups of machines and large groups of people.

Structure-Preserving Visualization of Scalar Fields
Wednesday, Sept. 28, 10:30-11:30 am, 50F-1647
Carlos D. Correa, Lawrence Livermore National Laboratory

In this talk, I will present topological spines—a new visual representation that preserves the topological and geometric structure of a scalar field. This representation encodes the spatial relationships of the extrema of a scalar field together with the local volume and nesting structure of the surrounding contours. Unlike other topological representations, such as contour trees, our approach preserves the local geometric structure of the scalar field, including structural cycles that are useful for exposing symmetries in the data. For high-dimensional functions, finding such a structure from sparse samples becomes increasingly difficult. I will also describe a number of approaches to improve the robustness of topological decompositions based on geometric neighborhood graphs.

Link of the Week: InfiniBand Architecture Impact on CPU Utilization

Joseph Yaworski writes in Scientific Computing:

High performance computing applications require a tuned and efficient performance from the CPU, interconnect, MPI and communication library to achieve optimal performance. Conventional wisdom would have you believe that the host channel adapters (HCAs) with offload capability, where processing power for running the communication library is on a card, would require less CPU utilization than HCAs using an on-load method, where the communication library is run on the host CPU, thus, allowing more CPU cycles for applications.

This view might not be correct. There are two very important factors to consider: first, the design and impact of the communication library when handling MPI application communications; and second, the impact of faster and higher core count processors in relation to the performance of the HCAs’ offload processing capability.

Today, two different InfiniBand architectures are available to run MPI computer applications. The first is where the HCA uses an offload processing approach and Verbs as the communication interface for MPI. The other uses on-load approach with a communication library called performance scaled messaging (PSM).

Read more.

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.