NERSCPowering Scientific Discovery for 50 Years


May 1, 2008


Harvey Wasserman from the Science-Driven System Architecture Team has been chosen as chair of the technical program for SC07, which will take place in Reno in November.


NERSC and CITRIS, the Center for Information Technology Research in the Interest of Society, co-hosted two meetings at UC Berkeley this month that focused on cognitive computing and online search technologies. NERSC Director Horst Simon gave the opening remark in both meetings, which attracted a total of 400 people from around the world.

The first meeting, “Cognitive Computing, a Multi-Disciplinary Synthesis of Neuroscience, Computer Science, Mathematics, Cognitive Neuroscience, and Information Theory” examined how scientists and engineers can build computer systems to become more flexible and adaptable, much like how the human brain and nervous system work.

The second meeting, “The Future of Search,” brought together UC Berkeley scientists and industry leaders, such as Yahoo, Microsoft and Google, to set search goals for building next-generation search engines. During the meeting, attendees discussed how to reach the ultimate goal of creating search engines with cognitive and natural language capabilities.

More information about the cognitive computing meeting can be found at . Learn more about the future of search at


Michael Resch, the head of the High Performance Computing Center Stuttgart (HLRS) visited NERSC last month, a stop-over to discuss issues facing both supercomputer centers. Resch spoke with NERSC staff about plans for upgrading systems and supporting scientists. They also chatted about operating energy-saving, large-scale supercomputers.

HLRS is part of the University of  Stuttgart in Germany. Resch, who serves as the director at the university’s Department of High Performance Computing, also discussed a plan by the president of the university, Wolfram Ressel, to visit NERSC in October.

“I’ve known NERSC and Horst Simon for a long time, and I was especially impressed that NERSC is able to keep a leading position in HPC over such a long time,” Resch said.


Several people recently joined  NERSC as full-time staff members,  serving as experts in areas such as  storage, system infrastructure, network and security. Here is a short  introduction of the new team members:  

Michael Welcome, who is now a  member of the Mass Storage Group, has over 23 years  experience in the field  of computing and computational science. For  the past seven years,  Welcome has worked  in the Future  Technology Group  within the  Computational Research Division of  Berkeley Lab, doing both application performance analysis and low-level network  and interconnect programming. His recent projects include porting  GASNet, the Berkeley UPC runtime  system, to the Cray XT3/XT4, as well  as porting and optimizing Miranda, a  Lawrence Livermore National  Laboratory (LLNL) hydrodynamics  code, to the LLNL BlueGene system. 

Welcome also analyzed the performance of a CCSE adaptive mesh refinement code on a collection of high-end  systems, including the Cray X1 at Oak  Ridge National Laboratory, IBM Power 5 at NERSC and the NASA Columbia  system.  

Previously, Welcome had worked in  the NERSC Computational Systems  Group, managing the Cray T3E and  the J90 cluster as well as writing and  maintaining the client-side accounting  and account management  software. Prior to this, Welcome  worked at LLNL from 1983-1996  developing adaptive mesh refinement  codes for both compressible and  incompressible fluid dynamics.

Jason Lee, who  just joined the  Networking,  Security, Servers  and Workstations  Group, holds a master’s degree in computer science from  San Francisco State  University. Lee has  worked at Berkeley Lab for the last 15  years, working on projects that range  from developing a parallel storage system to grid architecture in the Global  Grid Forum. More recently, he worked on Bro, a  network intrusion detection system  developed by the Lab and the  International Computer Science  Institute. Lee was involved in the early  quality-of-service (QoS) network tests  with Van Jacobson back in 1994,  when Jacobson was the head of the  Network Research Group at the Lab  and was known for his work on  improving IP network performance and  scaling.  

Lee worked on carrying out some of  the first gigabit test beds, including the  Multidimensional Applications and  Gigabit Internetwork Consortium  (MAGIC) project in 1993, Bay Area  Gigabit Network (BAGNet) in 1995 and National Transparent Optical  Network (NTON) in 1997. He also  worked with the Advanced Internet  Research group at the University of  Amsterdam on 10-Gbps transatlantic  links in 2002.

Tina Declerck recently retruned to the  NERSC’s Computational  Systems Group after  working there from 1997  through 2001. In her  previous tenure at  NERSC she managed  the PDSF and Cray J90  SV1 clusters. During her hiatus, she  worked at Sistina Software, the company  that developed the Global File System  (GFS) and was later bought by Red Hat. 

At Sistina, Declerck was in charge of  growing and managing the customer  support team.

At 3Ware, which manufactures SATA  RAID controllers, she defined new product features, acted as the primary interface between the field support and engineering teams, provided technical training  to resellers and worked on I/O performance and tuning.  

Now back at NERSC, Declerck is the  system lead for the SGI Altix named  DaVinci, and she provides secondary  support on Franklin, the Cray XT4, that  was recently installed.

Zhengji Zhao joined  the NERSC’s User  Services Group after  working for three years  as a postdoc in the  Computational Research  Division’s Scientific  Computing  Group (SCG) at  Berkeley Lab. In SCG, she worked on developing new methodologies for computational nanoscience:  the linear scaling 3D fragment (LS3DF)  method for large-scale electronic structure calculations and other methods.  

Zhao received her Ph.D. in computational physics from New York University  for developing the reduced density  matrix (RDM) method, a highly accurate  alternative to wavefunction-based computational chemistry methods. She  received a master’s degree in computer  science from the Courant Institute of  Mathematical Sciences, New York  University; a master’s degree in theoretical physics from Peking University in  Beijing, China; and a bachelor’s degree  in theoretical physics from Jilin  University in Changchun, China.

Akbar Mokhtarani  came to NERSC from  Berkeley Lab’s High  Energy and Nuclear  Physics Computing  Group, where he supported the BaBar detec- tor at the Stanford  Linear Accelerator  Mokhtarani  Center (SLAC) and the IceCube and  Majorana experiments. His responsibilities included migrating a database to  new software, working on a Java/C++  monitoring system and building and  packaging simulation software.  

Mokhtarani, who is in the Mass  Storage Group, is now applying his  database and software development  skills to the SciDAC Petascale Data  Storage Institute (PDSI), for which he  is gathering data and reporting on storage and file system reliability, and  working on I/O benchmarking of selected petascale applications. He also  works closely with NERSC’s Science- Driven System Architecture Group and  the team in charge of the Global File  System.

About NERSC and Berkeley Lab
The National Energy Research Scientific Computing Center (NERSC) is a U.S. Department of Energy Office of Science User Facility that serves as the primary high performance computing center for scientific research sponsored by the Office of Science. Located at Lawrence Berkeley National Laboratory, NERSC serves almost 10,000 scientists at national laboratories and universities researching a wide range of problems in climate, fusion energy, materials science, physics, chemistry, computational biology, and other disciplines. Berkeley Lab is a DOE national laboratory located in Berkeley, California. It conducts unclassified scientific research and is managed by the University of California for the U.S. Department of Energy. »Learn more about computing sciences at Berkeley Lab.