Computing Sciences Reorganizes to Better Align R&D Programs in High-Performance Computing and Networking Technology

We are now at the eighteen month anniversary of our inception as Computing Sciences. Our progress to date has been remarkable, and we have already earned a reputation for success and accomplishment in DOE. Computing Sciences has served as a catalyst, positively impacting both DOE national programs and the Berkeley Laboratory. New opportunities for scientific programs are being pursued through Grand Challenges, LDRDs, and the DOE 2000 initiative, combined with new collaborations with the Advanced Light Source, High Energy Physics, Nuclear Science, and the Human Genome project.

At this stage in our experience, we have taken the opportunity to evaluate our organizational structure to ensure its alignment with the program requirements of DOE and the needs of Berkeley Lab. We have identified two major objectives that would benefit from a restructuring and the resulting focus of our efforts:

1. The Berkeley Laboratory computing environment must be better defined and supported. Our ultimate goal is the integration of computing throughout the Laboratory's programs, making Berkeley Lab the computational laboratory of choice in the DOE community;
2. DOE programmatic requirements necessitate the coherent integration of our national facilities and computer science research. Given a declining Federal budget and increased competition for research funding, it is critical that Computing Sciences maintains a close alignment with the MICS program office so that we can best ensure our continued funding.

With these goals in mind, we are undertaking some changes. Specifically, the R&D programs are being realigned such that the major thrust of the ICSD R&D program will be in the areas of networking and collaborative technologies, while the major thrust of the NERSC R&D program will be in high performance computing including data management and data intensive computing. In addition, we are consolidating and strengthening our support activities for Berkeley Lab Computing.

Information and Computing Sciences Division

We are fully committed to positioning Berkeley Lab as the computational laboratory of choice in the DOE community. Thus, we are forming the Computing Infrastructure Support Department (CIS) to significantly improve support for Berkeley Lab Computing, integrating the associated UNIX and VMS support groups with desktop workstation support already in ICSD. Stu Loken, Division Director, has asked Eric Hibbard to serve as Department Head of this important start-up activity.

TEID, under Acting Department Head, Dennis Hall, remains within ICSD, as does the Information Systems and Services Department (ISS) under Carl Eben. Additionally, Stu Loken serves as the principal liaison with Berkeley Lab Operations for all Computing Sciences infrastructure activities, including LBLnet, Telephone Services and Visualization, and has responsibility for the annual prioritization of associated funding requests. Stu has asked Carl Eben to serve as Deputy Division Head with specific responsibility for ISS and CIS.

Under Stu's leadership, the DOE 2000 Initiative is achieving early national success with major Berkeley Lab participation. These activities, and those of the current Imaging and Distributed Computing Group that support the creation of scientific collaboratory environments, will form a new DOE 2000 group with Bill Johnston as Group Leader. This, combined with the Comprehensive Epidemiologic Data Resource (CEDR), HENP Computational Sciences Group, Networking Research Group, and Visualization serve as the mainstream R&D components of ICSD. The HENP Computational Sciences Group will develop and advance computational approaches and tools for use in high energy physics and nuclear physics applications.

In addition, Stu has agreed to continue his half-time assignment in Washington providing expert support to the MICS staff in the development and deployment of new programs.

National Energy Research Scientific Computing (NERSC)

Division
NERSC's mission is to accelerate the pace of scientific discovery in the DOE ER community. From the beginning, the design for accomplishing this goal at the Berkeley Lab has been to provide leading edge high performance computing services while introducing the best new computing science tools and innovations. This reorganization will enable us to achieve a strengthened programmatic focus by bringing high performance computing services and associated research and development together.

Bill Kramer continues to serve as High Performance Computing Department (HPCD) Head for NERSC providing advanced technology and support for the DOE scientific mission. In addition, Bill will be the acting group leader of the newly formed HENP Computing Systems group, which will serve as the foundation for our planned future initiatives in this arena. The PDSF will form the core of this new group within the HPCD. This new group will create an LBNL focus for significant new opportunities to develop, integrate, and support physics computing systems dedicated to the international HENP community.

Within the NERSC Division, the recently formed High Performance Computing Research Department (HPCRD) will integrate the efforts of the following groups: Bioinformatics, Data Intensive Computing, Future Technologies, Scientific Computing, Scientific Data Management Research, and Visualization. Those projects of the current Imaging and Distributed Computing Group related to high performance computing will form the Data Intensive Computing group, with Brian Tierney as its leader. The Collaborative Management Environment and Distributed Notebooks efforts of the Data Management group will be matrixed to ICSD's DOE 2000 Group.

Horst Simon, NERSC Division Director, will serve as Acting Head of the research efforts as he leads the search efforts for a permanent Head of the High Performance Computing Research Department. He has asked Bill Kramer to serve as NERSC Division Deputy Head, with Bill's major focus on NERSC program deliverables, including the procurement of the next major capability system.

Bill Kramer will also serve as the Computing Sciences focal point for building the computing support for the Human Genome project.