InTheLoop | 02.09.2009
The weekly newsletter for Berkeley Lab Computing Sciences
February 9, 2009
Berkeley Lab Issues Call for LDRD Proposals
Interim Lab Director Paul Alivisatos has issued a call for proposals (CFP) for the FY 2010 Laboratory Directed Research and Development (LDRD) program. The LDRD program constitutes one of the principal means to seed innovative science and new research directions.
For the FY 2010 cycle, there will be a mutually exclusive two-track review process. The first track, similar to prior years, will be proposals reviewed as Divisional and/or Laboratory-wide. The second track will be for “Discovery” proposals. Up to ~$1.0M of the FY 2010 budget will be reserved for funding a set of smaller proposals with higher scientific risk and potential payoff. They will be reviewed and funding recommendations made by external scientific reviewers.
The LDRD CFP and downloadable forms are available on the web at http://www.lbl.gov/dir/LDRD/cfp.
HSS Audit Wraps Up
The DOE Integrated Safety Management HSS Review wrapped up its field visit last Thursday. Berkeley Lab received some positive feedback and also indications that some areas need improvement. In the debrief by auditor Tom Staker, preliminary observations were noted. Notes from this event are available at https://www.lbl.gov/ehs/ism/2009/outbrief.html. The Lab as a whole seemed to have a good safety culture; some areas such as the JHA, chemical management, and safety documents will need work.
CRD, NERSC Staff Will Speak at IPDPS 2009 Symposium
Staff from Berkeley Lab’s Computational Research and NERSC divisions will give five presentations during the IEEE International Parallel & Distributed Processing Symposium to be held May 25–29 in Rome, Italy. IPDPS is an international forum for engineers and scientists from around the world to present their latest research findings in all aspects of parallel computation. In addition to technical sessions of submitted paper presentations, the meeting offers workshops, tutorials, and commercial presentations and exhibits.
“This is one of the most competitive IEEE conferences, and the fact that we have so many papers speaks for the quality of our computer science program,” said Horst Simon, applications chair for the conference and Associate Lab Director for Computing Sciences at LBNL.
LBNL contributors are:
- Lenny Oliker of the Future Technologies Group, one of three invited keynote speakers, will give his talk “Green Flash: Designing an Energy Efficient Climate Supercomputer” on Thursday, May 28.
- Ekow Otoo, Doron Rotem, and Shih-Chiang Tsao of the Scientific Data Management Group will present their paper on “Analysis of Trade-Off between Power Saving and Response Time in Disk Storage Systems” as part of the Fifth Workshop on High-Performance, Power-Aware Computing on Monday, May 25.
- Kamesh Madduri of the Scientific Data Management Group, along with David A. Bader of the Georgia Institute of Technology, will present their paper on “Compact Graph Representations and Parallel Connectivity Algorithms for Massive Dynamic Network Analysis” as part of the Graph and String Applications session on Tuesday, May 26.
- Brian Van Straalen, Terry Ligocki, and Noel Keen of the Applied Numerical Algorithms Group, John Shalf of NERSC, and Woo-Sun Yang of Cray Inc. will present “Scaling Challenges for Massively Parallel AMR Applications” during the Scientific Applications Session on Thursday, May 28.
- Rajesh Nishtala, Paul Hargrove, and Dan Bonachea of the Future Technologies Group and NERSC Director Kathy Yelick will present “Scaling Communication Intensive Applications on BlueGene/P Using One-Sided Communication and Overlap” during the Communications sSystems Session on Thursday, May 28.
David Bailey Will Join AMS Notices Editorial Board
CRD Chief Technologist David Bailey, who is well known as an author and pioneer in the field of experimental mathematics, has been invited to join the editorial board of Notices of the American Mathematical Society, the monthly membership journal of the AMS. His three-year term will begin with the January 2010 issue. As a board member, Bailey will suggest topics and authors and will offer advice on editorial issues.
CRD Report Highlights Environmental Database
The latest edition of CRD Report is now online. Highlights of this issue include:
- New Tools Mobilize Local Data to Study Global Environmental Issues
- David Patterson: The Future of Computer Architecture
- On Tour: CRD Researchers Share Their Expertise at Home and Abroad
- ESnet Upgrades Science Network
Application Deadline Today for Tapia and Hopper Conferences
The Computing Sciences Diversity Working Group is working on three initiatives this year:
- Sponsorship of employee travel to attend the Tapia Celebration of Diversity in Computing and the Grace Hopper Celebration of Women in Computing conferences. To apply for funding, send your name, group, and department, the conference name, and a paragraph describing why you would like to attend the conference to firstname.lastname@example.org. The deadline for submitting the above information is today, February 9, 2009.
- Sponsorship of an employee mentoring/networking program, modeled after a very successful program under way at LLNL. Details of this program will be announced soon.
- Sponsorship of minority summer students. If you are interested in hiring a summer student from an under-represented population who would be eligible for funding, please let us know at email@example.com.
NERSC Has Opening for a Computer Systems Engineer
The NERSC Science-Driven Systems Architecture team has an opening for a Computer Systems Engineer. This position will contribute to an ongoing team effort to develop a complete understanding of the issues that contribute to optimal application and computer system performance on extreme-scale advanced architectures. The engineer will assist NERSC in evaluating existing and emerging HPC systems by analyzing the performance characteristics of leading-edge DOE Office of Science application codes. The successful applicant will require knowledge of computer architecture, with a particular focus on understanding the implications of multicore computing technology on the scalability and programmability of future scientific computing applications. See the job details here.
The Lab’s Employee Referral Incentive Program (ERIP) awards $1,000 (net) to employees whose referral of an external candidate leads to a successful hire.
DOE Is Accepting Nominations for Lawrence Awards
The Department of Energy invites nominations for the Ernest Orlando Lawrence Awards, which honor US. scientists and engineers at mid-career for exceptional contributions in research and development supporting the Department of Energy and its mission to advance the national, economic, and energy security of the United States. The Award consists of a citation signed by the Secretary of Energy, a gold medal, and a $50,000 honorarium.
An award is given in each of the following fields: Chemistry, Materials Research, Environmental Science and Technology, Life Sciences (including Medicine), Nuclear Technologies (Fission and Fusion), National Security and Nonproliferation, and High Energy and Nuclear Physics. Nominations for the 2009 Lawrence Awards must be submitted by April 1, 2009. The Lawrence Awards website describes the nomination guidelines and process.
Workshop on Real-Time Astronomy to Be Held in Santa Cruz
Registration is now open for “Hotwiring the Transient Universe II: Real-Time Astronomy,” to be held at the UCO/Lick Observatory at UC Santa Cruz, April 26–30, 2009.
The astronomical time domain ranges from solar physics and solar system objects to objects and processes at galactic and cosmological distances. Transients arrive via electromagnetic radiation, gravitational waves, neutrinos and other particles. Discoveries are made via spacecraft and by ground-based surveys, through automatic pipelines and the Virtual Observatory, with robotic telescopes and by the human eye. A strong interdisciplinary agenda will engage the emerging information infrastructure for astronomical transient events and their rapid follow-up. Session themes will include:
- Connecting a robotic telescope to an event stream
- Real-time mining of event streams
- Event streams: Catalina, OGLE, MOA
- BoF for event providers: LOFAR, Skymapper, GCN, etc.
- Event authoring and real-time dissemination.
Green Collar Economy Is Topic of BERC Lecture
The Berkeley Energy and Resources Collaborative (BERC) is presenting the second annual BERC Lecture, “The Green Collar Economy,” at 6:15 pm on Wednesday, February 11, at the Bancroft Hotel, 2680 Bancroft Way, Berkeley. The speaker will be Van Jones, founding president of Green For All, senior fellow with the Center for American Progress, and founder of the Ella Baker Center for Human Rights. Here is the abstract:
As climate chaos and soaring fuel costs batter the U.S. economy, Van Jones brings a unique perspective to the color green. He argues that a green economy that rescues the polar bears can also rescue the U.S. economy — putting America back to work; driving down energy and food prices; giving green collar jobs and new chances to underserved youth; and forever ending the need for oil wars and resource wars.
This Week’s Computing Sciences Seminars
SGI Vis/Graphics Roadmap and UltraViolet
Monday, February 9, 10:00 am, 50F-1647
SGI staff will talk about their vis/graphics roadmap as well as UltraViolet, which is their future HPC architecture.
Multi-Touch Interfaces: Their History and Their Future
Tuesday, February 10, 10:00–11:30 am, Sibley Auditorium, Bechtel Engineering Center, UC Berkeley
Bill Buxton, Microsoft Research
Regents’ Lecturer Bill Buxton began his career in the computer music field and early on stressed the important role of human computer interaction in musical creation and performance. His work in music spread to graphics and, in fact, to the entire field of human-computer interaction design. His recent book Sketching User Experiences, published by Microsoft Press, has received wide ranging critical acclaim. This work has been reviewed on BusinessWeek.com and strategy-business.com — the latter states that the book “has earned its spot as the best innovation book of this year.”
Program Analysis 2.0
Tuesday, February 10, 1:00–2:00 pm, 310 Soda Hall, UC Berkeley
Thomas Ball, Microsoft Research
Microsoft Research’s efforts in program analysis began with detecting defects in C and C++ programs with tools such as ESP, PREfast, PREfix, and SLAM, all of which have been widely deployed internally. The PREfast and SLAM technologies also were incorporated into shipping Microsoft products (Visual Studio and the Driver Development Kit, respectively).
In the last few years, we have been working on program analysis tools for .NET, focusing on: (1) code contracts and verifying code against contracts; (2) automatic test generation; (3) checking for concurrency defects. These tools are based on advances in abstract interpretation over linear inequalities, symbolic execution of object-oriented programs, efficient and precise automatic theorem proving, and direct model checking of concurrent systems. All of the above tools are available for both academic and commercial use, in partnership with Visual Studio.
Future Directions in High Performance Computing (HPC) 2009–2018
Wednesday, February 11, 11:10 am–noon, 380 Soda Hall, UC Berkeley
Horst D. Simon, ALD for Computing Sciences, LBNL
Since about 2004 there has been a fundamental transition taking place in computing. The microprocessors’ clock speed improvements have leveled off, and future performance increases from processors will be realized through multi-core and many-core chips.
This change in the basic building blocks for HPC has opened up the architecture discussion for future HPC platforms, and in 2009 we see vigorous experimentation with accelerators, GPUs, FPGAs, and embedded technology. HPC has not seen such a variety of new technology being explored since the early 1990s. In my talk I will explore what the multi-core revolution will mean for the future of HPC. I will use the very successful model of the HPC ecosystem and its important elements — economic driver, system architecture, and programming model — to explain how I think HPC will develop in the next five years. Several of the projects that are currently going on in Berkeley will be discussed in detail, since they will provide tools for the future productive use of supercomputers. They include auto-tuning, PGAS languages, and most importantly the Green Flash project, that attempts to find a new solution for energy efficient computing in the future.
(How) Can Programmers Conquer the Multicore Menace?
Thursday, February 12, 11:00 am–12:30 pm, 430 Soda Hall, UC Berkeley
Saman Amarasinghe, CSAIL/EECS, MIT
The era of exponential growth of processor performance, transparent to the programmer, is over. Multicores put the onus of taking advantage of Moore’s law on the programmers. While architects have known how to build parallel processors for over a half a century, programmers’ inability to create scalable parallel programs has been the main stumbling block in mainstream use of parallelism. In the first part of the talk I will discuss the path to multicores, address why scalable parallel programming has been such a difficult problem to solve, and speculate on our ability to crack it this time around. Next I will discuss two possible technologies, Petabricks and Kendo, which can alleviate part of the heavy burden of parallel programming we are placing on programmers.
For a multicore application to keep up with the exponential growth of Moore’s Law requires the algorithms in the application to efficiently scale from small to large numbers of cores. It is observed that for a given problem, different algorithms provide the best solution at different levels of parallelism. However, currently there is no simple way for the programmer to express or the compiler to take advantage of all the available algorithmic choices for a problem. PetaBricks is a new implicitly parallel language and compiler where having multiple implementations of multiple algorithms to solve a problem is the natural way of programming. The PetaBricks compiler autotunes programs by making the best fine-grained algorithmic choices. Choices also include different automatic parallelization techniques, data distributions, algorithmic parameters, transformations, and blocking.
Non-determinism, inherent in parallel applications, causes significant challenges for parallel programmers by hindering their ability to create parallel applications with repeatable results. As a consequence, parallel applications are significantly harder to debug, test, and maintain than sequential programs. Kendo is a new software-only system that provides deterministic multithreading of parallel applications. Kendo enforces a deterministic interleaving of lock acquisitions and specially declared non-protected reads through a novel dynamically load-balanced deterministic scheduling algorithm. Kendo can run on today’s commodity hardware while incurring only a modest performance cost 16% on SPLASH-2 benchmarks when running on four processors.
Efficient and Robust Multipurpose Linear Solvers
Friday, February 13, 10:00–11:30 am, 50F-1647
Haim Avron, Tel-Aviv University, Israel (Alvarez Fellowship candidate)
Solving linear equations is an important part of many scientific computing tasks. A special-purpose solver can often be the most efficient, but in many cases no special-purpose solver exists, and designing a new one is too expensive or would take too long. Such problems benefit from robust and efficient multipurpose solvers. In my talk I will present robust and efficient multipurpose linear solvers I developed during my studies.
The first part of my talk will focus on solving symmetric positive systems. Incomplete LDLT factorizations are often effective as preconditioners, but they sometimes produce an indefinite preconditioner even when the input matrix is symmetric positive definite. The two most popular iterative solvers for symmetric systems, MINRES and CG, cannot use such preconditioners; they require a positive definite preconditioner. I will present two new Krylov-subspace solvers, a variant of MINRES and a variant of CG, both of which can be preconditioned using any non-singular symmetric matrix as long as the original system is positive definite.
The second part of my talk describe a solver for rank-deficient overdetermined least-squares problems. It perturbs a sparse QR; that is, we do not factor the original coefficient matrix but a perturbed one. This allows our solver to solve rank deficient problems without column pivoting. It is efficient and easy to parallelize.
I will also briefly discuss two other solvers. One is a solver for finite element discretizations of scalar elliptic PDEs that uses a combinatorial preconditioner. The second is a parallel direct solver for sparse unsymmetric linear systems.
Link of the Week: Women in Computing—Take 2
In the February 2009 issue of Communications of the ACM, Maria Klawe, Telle Whitney, and Caroline Simard write:
“Women in Computing: Where Are We Now?”—an article by Maria Klawe and Nancy Leveson in the January 1995 issue of Communications—addressed women’s representation at the time, as undergraduate and graduate students and in the work force, in computing fields. That article, part of the issue’s special section on Women and Computing, described successful activities and offered recommendations for future programs.
In this article, 14 years later, we assess the changes that have since occurred, including both positive and negative trends; we present strategies shown to be successful for the recruitment, retention, and advancement of women in computing; and we explore promising new initiatives for further increasing women’s participation.
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 7,000-plus 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 Department of Energy 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.