High-Performance Computing and Communications

Among supercomputers in an academic setting, HPCC's new supercomputer cluster is the 6th fastest in the United States. Among all supercomputers in the world, it is ranked 71st. It claimed this distinction by achieving a benchmark in fall 2009 of 72.05 teraflops, or 72.05 trillion floating-point calculations per second, on its 1280-node, 10-gigabit backbone cluster.

HPCC comprises a diverse mix of computing and data resources. Two Linux clusters constitute the principal computing resource. In addition, HPCC has a central facility that provides more than 400 terabytes of combined disk storage and potential access to nearly a petabyte of tape storage, as well as a Condor cluster that uses spare cycles on UNIX workstations and PCs in USC's general-access computing rooms.

Upcoming HPCC Downtimes

1) Scheduled Downtime for 2-GB Cluster (March 1 - 5)

From 10:00 a.m. on Monday, March 1, 2010, until 10:00 a.m. on Friday, March 5, 2010, we will be decommissioning nodes on our older 2-GB, low-latency bandwidth cluster.

The following nodes will be unavailable during this period: 138 of the V20, single-core, dual-processor nodes; 120 of the V20, dual-core, dual-processor nodes; and 40 dual-core, dual-processor sc1425 nodes.

An additional 120 v20, dual-core, dual-processor nodes will be relocated and therefore unavailable at this time.

The remaining nodes on the 2-GB cluster, and the entire 10 GB cluster, will continue to be available.

2) Scheduled Downtime for the 2-GB and 10-GB Cluster (March 22)

From 10:00 a.m. on Monday, March 22, until 5:00 p.m. on Monday, March 22, all head nodes, compute cluster nodes, almaak, and file servers on the 2-GB and 10-GB cluster will be unavailable. During this time, security and operating system patches will be applied.

Services will be restored as they become available.

3) Scheduled Downtime for 10-GB Cluster (March 22 - 29)

From 10:00 a.m. on Monday, March 22, until 10:00 a.m. on Monday, March 29, the entire 10-GB low-latency bandwidth Linux cluster will be unavailable due to installation of 180 64-bit, quad-core, dual-processor nodes.

During this downtime, the 2-GB cluster will remain available.

HPCC Research		HPCC System Updates
Thieo Hogen-Esch Elucidation of the mechanisms of the Yamamoto polymerizations More... James Knowles and Peter Laird Analysis of DNA sequences generated with next-gen technology More... John O'Brien Active and passive nanophtonic integrated circuit components More... Thomas Jordan Using seismic tomography to help reduce earthquake risks More... Carl Kesselman Middleware guru and grid pioneer More... Kevin Knight and Daniel Marcu Statistical approach to natural language translation More... Aiichiro Nakano and Ashish Sharma Large-scale scientific visualization More... Priya Vashishta Multi-teraflop supercomputing More... Arieh Warshel Computer simulation of biological processes More... Richard Weinberg Computer animation and visual effects More...		Node Upgrade On Monday, September 28th, 384 IBM iDataPlex dx340 systems were added to the cluster, replacing 342 Sun V60x. Each node is a quad core/dual processor 2.33gHz, with 16GB of memory.