InTheLoop | 11.15.2010
November 15, 2010
Hopper Breaks Petaflop/s Barrier, Ranks No. 5 in TOP500 List
Hopper, a Cray XE6 system at NERSC, broke the petaflop/s barrier by achieving 1.05 petaflop/s running Linpack, the TOP500 benchmark application. This makes Hopper the second most powerful system in the U.S., the fifth fastest in the world, and only the third U.S. machine to achieve petaflop/s performance.
The 36th edition of the TOP500 list of the world’s most powerful supercomputers has just been published online. Five of the top ten computers are in the U.S., and four of those are funded by the Department of Energy.
Berkeley Lab Showcases Expertise at SC10 in New Orleans
The SC10 conference is under way all week in New Orleans, expanding the frontiers of HPC by creating a potent platform for demonstrating, debating and discovering innovative, cutting-edge advances in computation, networking, storage, and analysis through an exceptional technical program, an expanded exhibits area, an exciting education program and much more.
At Booth 2448, Berkeley Lab is showcasing our valuable computing, networking and scientific resources—our world-class roster of recognized experts. Many of our staff members will be holding “office hours” and talks in our booth. We are also showcasing our renowned scientific visualizations in 3D and on our globe. Click here for a complete listing of events featuring LBNL staff, including technical papers, talks, roundtable discussions, tutorials, workshops, demos, posters, and birds-of-a-feather sessions.
SC10 has partnered with Boopsie to develop a free mobile phone app to help attendees navigate the extensive program and provide customized conference information. Attendees can download the app at http://sc10.boopsie.com. You can then:
- Customize your schedule with sessions, speakers and events you want to attend
- Plot out the exhibit floor with the booths you want to visit
- Get New Orleans hotel and restaurant info
- Follow SC10 via social media
Phil Colella to Deliver Invited Talk on Climate at SC10
Phil Colella, leader of CRD’s Applied Numerical Algorithms Group, will be one of the invited speakers in the Masterworks program at the SC10 conference being held this week in New Orleans. Colella will discuss “High-End Computing and Climate Modeling: Future Trends and Prospects” during the Big Science, Big Data session beginning at 1:30 p.m. Tuesday. Here’s the abstract for his talk:
Over the past few years, there has been considerable discussion of the change in high-end computing, due to the change in the way increased processor performance will be obtained: heterogeneous processors with more cores per chip, deeper and more complex memory and communications hierarchies, and fewer bytes per flop. At the same time, the aggregate floating-point performance at the high end will continue to increase, to the point that we can expect exascale machines by the end of the decade. In this talk, we will discuss some of the consequences of these trends for scientific applications from a mathematical algorithm and software standpoint. We will use the specific example of climate modeling as a focus, based on discussions that have been going on in that community for the past two years.
ESnet Continues Tradition of Participation in SCinet
What does it take to provide state-of-the-art networking to a conference like SC10? ESnet’s Jon Dugan gives an insider’s perspective in the latest Network Matters blog, “Autumn means SC10,” where he talks about the people who make SCinet, the annual conference’s temporary network, possible.
Nominations Sought for BER Director
Patricia Dehmer, Deputy Director for Science Programs in the DOE Office of Science, is requesting help in identifying candidates for the position of Director of the Office of Biological and Environmental Research (BER). The BER science portfolio includes research programs and scientific user facilities that address some of today’s most important and DOE mission-critical problems in biological, climatic, and environmental research.
Suggestions for candidates for this position, including self nominations, are welcomed. Each individual suggested will be sent a letter providing information on the position and how to apply, and will be encouraged to submit an application. Please send your suggestions to firstname.lastname@example.org, with an e-mail address for each candidate, by Monday, November 22.
NERSC, BDMTC Have Openings for Computer, Software Engineers
NERSC has an opening for a Technology Integration Engineer to help build an ecosystem to support data-intensive science. This includes investigating emerging technologies such as solid-state storage, MapReduce, and NoSQL frameworks, and evaluating how these technologies can help tackle scientists’ most demanding data challenges. See job details.
The Biological Data Management and Technology Center (BDMTC) in CRD has an opening for a Senior Software Engineer for genomic data management and software development projects. This engineer will coordinate the development of a Human Microbiome Project specific metagenome data management and analysis system; work on development of components for a software system for supporting managing genome sequencing projects at the Joint Genome Institute (JGI); and assist the head of BDMTC in examining the genome data processing and analysis activities across JGI’s scientific programs with the goal of improving their efficiency and performance. See job deatails.
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.
Big Ideas at Berkeley: Student-Led Innovation Is Changing the World
Since its founding in 2006, UC Berkeley’s annual “Big Ideas” prize competition has inspired innovative and high-impact student projects aimed at solving the most pressing problems facing the world. By providing concrete support, visibility, and review, Big Ideas challenges students to set ambitious goals for current and future endeavors.
On Thursday, November 18, to launch the 2011 contest, John Seely Brown, former Chief Scientist of Xerox and the director of its Palo Alto Research Center (PARC), will speak on “Student-Led Innovation Is Changing the World” in the Banatao Auditorium of Sutardja Dai Hall at 12:30 pm. Refreshments will be served. To learn about previous winners, go here.
This Week’s Computing Sciences Seminars
Par Lab Seminar: An Established Solution for Programming Productivity and Software Performance on Multiprocessor, Multicore and Manycore Systems
Tuesday, Nov. 16, 11 am–12:30 pm, 438 Soda Hall (Wozniak Lounge), UC Berkeley
William Lundgren, Gedae, Inc.
The fundamental problem in software development has been stated by many people including Burton Smith (Technical Fellow at Microsoft Corp.) who says, “We have to reinvent computing, and get away from the fundamental premises we inherited from [traditional compilers].” Gedae is an established technology for automatically creating highly efficient applications for multiprocessor architectures, with notable successes in high performance embedded computing and in benchmarking the Cell/BE architecture. Gedae’s technology consists of four major components: a language, an architectural modeling language, an OS and a compiler. These four components were designed together as interrelated parts in a single, all-in-one solution. The key innovation of the solution is that it abandons the traditional software stack that is built on compilers for single cores and provides a new solution better suited to modern and future parallel systems. Gedae moves complexity typical found in a software or hardware stack to compilation, including management of memory and cache, shared data and state, concurrency control and load balancing, etc. This stackless approach is more extreme than one might expect. Gedae does not just replace middleware, it streamlines how processors are used and provides optimizations and verification not possible with other technology.
While traditional languages like C and Fortran are insufficient to address the problem with software development, contrary to what many people believe, a new parallel language is not required to solve the multicore programming problem. A language that purely specifies functionality with no implication on the way the software will be implemented can be used. If the language sufficiently exposes the parallelism, a compiler can be given maximal opportunity to optimize the software for the behavior and target. Gedae’s language is multilayered. The Gedae Data Flow language (Gedae DF) is an efficiency language that enables many of the compiler optimizations. A focus of Gedae’s research has been in extending Gedae DF to be general purpose. Provided on top of Gedae DF is a user-friendly productivity language named IdeaTM, acting as a thinking toy much in the same way Matlab is used.
The compiler technology is compatible with a wide range of current and future systems, including heterogeneous systems, emerging mobile devices and cloud servers. This breadth of support and forward compatibility is enabled by the hardware modeling language. The hardware model describes the details of the target hardware with information the compiler needs to construct the software, including number and type of CPUs, memory hierarchy, and interconnect. This model is input to the compiler along with the code and a description of how the code is mapped to the hardware model, and the compiler is tasked with converting those three components into highly efficient parallel code.
CITRIS Research Exchange: Furthering Math Education
Wednesday, Nov. 17, 12:00 pm, Banatao Auditorium, Sutardja Dai Hall, UC Berkeley
Deborah Zimmerman, Director of Production, Agile Mind
Live broadcast at mms://media.citris.berkeley.edu/webcast
As a nation, over half of our students fail algebra every year. Agile Mind was founded with the mission of changing what happens between educators and students in the classroom in ways that improve the quality of instruction of high school mathematics and science, especially in underserved areas. We use technology to provide broad and equitable access to best practices, enhance the classroom experience, support R&D at low cost, and provide an empirical basis for our services.
EECS Colloquium: for example...
Wednesday, Nov. 17, 4:00–5:00 pm, 306 Soda Hall (HP Auditorium), UC Berkeley
Scott Klemmer, Professor, Computer Science, Stanford University
In this talk, I will describe three research threads that investigate how software tools can increase the quality of people’s creative work — especially interface design and programming. The first explores leveraging online examples of creative work to empower more users to design new user interfaces and software programs, learners to acquire new skills, experts to be more creative, and programmers to engage in more design thinking. The second introduces techniques for designers to rapidly create novel user interfaces, explore more alternatives, and revise prototypes based on feedback. The third explores the psychological and social ingredients of design excellence — focusing on the role of alternatives and prototyping.
Link of the Week: What's Next in Science
“We’re going to see scientific results that are correct, that are predictive, but are without explanation. We may be able to do science without insight, and we may have to learn to live without it. Science will still progress, but computers will tell us things that are true, and we won’t understand them.”
That’s the prediction of Steven Strogatz, Professor of Applied Mathematics at Cornell University, and number ten in the New York Times text and audio feature Voices: What's Next in Science. Strogatz was referring to Eureqa, an “automated scientist” created by a Cornell engineer, Hod Lipson, and his students. In 2009, they reported that simply by observing a pendulum, Eureqa can rediscover some of Newton’s laws of physics. Eureqa is now looking for hidden patterns in the networks of proteins that break down food in cells.
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.