NERSC Systems

Hopper - Cray XT5

The system will arrive in two phases. The Phase I is an XT5 and will be delivered in 2009 and includes 668 compute nodes, each containing two 2.4 GHz AMD Opteron quad-core processors (5,344 total cores). The second stage, scheduled to arrive in 2010, will be combined with an upgraded Phase I to create a system with over 150,000 cores and a peak performance of 1.17 Pflop/s (118 Tflop/s sustained). The system is named in honor of American Computer Scientist Grace Hopper.

Franklin - Cray XT4

The NERSC Cray XT4 system, named Franklin, has 9,572 quad core compute nodes, i.e., a total of 38,288 processor cores available for scientific applications. Each compute node has a 2.3 GHz quad-core AMD Opteron processor and 8 GBytes of memory. The full system consists of 102 cabinets with 78 TBytes of aggregate memory. The theoretical peak performance of Franklin is about 355 TFlop/sec. The system is named in honor of Benjamin Franklin.

Carver - IBM iDataPlex

The NERSC IBM iDataPlex system, named Carver, has a total of 3200 computational cores with a peak performance of 34.2 TFlop/s. They are configured as 400 8-core nodes, each node having 24 GB of memory. The machine is named in honor of George Washington Carver, an American scientist, botanist, educator and inventor whose studies and teaching revolutionized agriculture in the Southern United States.

Euclid - Sun Firefox

Euclid is a Sun Microsystems Sunfire x4640 SMP. Its single node contains eight 6 core Opteron 2.6 GHz processors with all 48 cores sharing the same 512 GBytes of memory. Euclid's main purpose is to provide visualization and data analysis capabilities to the NERSC user community.

PDSF - Linux

NERSC operates PDSF, an approximately 1000-compute-core cluster, used by large scale Physics, High Energy Physics, Astrophysics and Nuclear Science investigations for detector simulation, data analysis, and software development. Storage needs are met by a heterogeneous mix of "disk vaults" and local storage that provide approximately 450 TB of disk space, as well as high-bandwidth connections to the NERSC Global Filesystem (NGF).

NGF - Global Filesystem

The NERSC Global Filesystem (NGF) is a large, shared filesystem that can be accessed from any of the compute platforms. This facilitates file sharing between platforms, as well as file sharing among NERSC users working on a common project. NGF is based on IBM's General Parallel File System (GPFS).

HPSS - Mass Storage

Archival mass storage is provided by the High Performance Storage System. This system has 100 TB of cache disk, 8 STK robots, and 44,000 tape slots for a maximum capacity of about 44 PB. HPSS archives 2.6 petabytes (PB) of data in 53 million files and sustains an average transfer rate of more than 100 MB/s, 24 hours per day, with peaks to 450 MB/s.

Data Transfer Nodes

The data transfer nodes are NERSC servers dedicated to performing transfers between NERSC data storage resources such as HPSS and the NERSC Global Filesystem (NGF), and the storage resources of other sites.

Network

NERSC offers end-to-end data transfer optimization and other network services. Access to NERSC from anywhere in the U.S. or the world is available through ESnet.

SVN/CVS Server

NERSC hosts Subversion (SVN) and CVS repositories for NERSC projects. Access is by SSH only.

Retired Systems