TITLE:

RECENT PROGRESS IN COMPUTATIONAL SCIENCE AT NERSC (No. 6)

DATE:

Friday, September 5th, 2003

TIME:

3:30 PM

LOCATION:

GMCS 214

SPEAKER:

Horst D. Simon, National Energy Research Scientific Computing (NERSC) Center, Lawrence Berkeley National Lab

ABSTRACT:

The NERSC (National Energy Research Scientific Computing) Center at Lawrence Berkeley National Laboratory is one of the largest open unclassified supercomputing centers in the world, providing capability resources for computational projects in areas such as astrophysics, climate modeling, combustion, fusion, and nano sciences. In early 2003 NERSC upgraded is computational facility to a 6,656 processor IBM SP with 10Tflop/s peak, which places it at the #5 rank of the TOP500 list.

NERSC also manages archival storage with 8.8 Petabyte capacity. Scientists have access to NERSC via OC-192 connections provided by ESnet. NERSC is working towards integrating these resources into a Unified Science Environment, providing grid services in particular for climate and high energy physics applications.

In this presentation I will first introduce NERSC in some detail. Then I will describe several computational projects, which demonstrate that the combination of computing, storage, and grid resources at NERSC provide an excellent environment for computational science. In particular, I want to show how the combination of a large computational resource at NERSC together with the work in the computer science and applied mathematics is beginning to impact the applications projects. Increasingly we can see today a change in simulation science. Computational science is moving to a new level of large scale, geographically dispersed collaborations of applications scientists with computer scientists and applied mathematicians, and this trend is evident at NERSC.

I will conclude with an outlook about the future of the rapidly changing environment for high end computing in the US. Several new initiatives are in the making, which indicate a renewed strong interest in all aspects of high performance computing.

HOST:

Jose Castillo

DOWNLOAD: