High Performance Computing

Summary: High Performance Computing (HPC) resources are available to UCI researchers. HPC resources include facilities at UCI ("HPC" and "Green Planet"), within UC ("ShaRCS Clusters"), and through national super-computing services such the San Diego Supercomputer Center (SDSC) at UCSD, and at other affiliated universities.

"HPC" Computer

Also at UCI, is the "HPC" Beowulf cluster is used for parallel processing and is available campus-wide. New faculty group 'Partners' to the system are now being encouraged to join the current faculty consortium. All UCI users can make use of the 'publicly' available compute nodes on this cluster.

Green Planet Cluster

Also at UCI, this cluster is owned and operated by the School of Physical Sciences and is used mostly for parallel computing. Faculty groups outside Physical Science are asked to contact Physical Sciences Computing Group for more information on the system including how to become a faculty partner. There is no access for the general UCI user to the system.

ShaRCS UC Pilot Research Computing Cluster

UC supported Pilot Research Computing clusters housed at both the San Diego Super Computing Center and the Lawrence Berkeley Laboratory. The project is studying the ability of a shared, remote, computing cluster to provide parallel processing computing services for the entire UC system.

Proposed Future Business Model for Further Development of ShaRCS

This document describes a proposed business plan to expand the current ShaRCS clusters with proposed joint PI and UCI funded compute servers.

Presentation to UC Vice Chancellor's for Research on ShaRCS Project (Powerpoint Slides)

This presentation to the UC VCRs group defines the current and proposed state of the ShaRCS system. Presentation given 5-23-2011 includes some updates to the business model which now include 'standard' and 'higher performance' compute nodes and costs.

San Diego Supercomputer Center (SDSC) Resources

Access to the SDSC machines does not require a proposal unless significant computational resources are required. Additional HPC resources are available through NSF, or at other universities, though access to those machines usually requires a formal proposal to the NSF to demonstrate need.

HPC Mailing List:

OIT is compiling a mailing list and directory of UCI faculty and researchers who use, own, or operate shared research computing clusters on campus. We define a cluster as being more than four servers or 'nodes' with at least 8 CPU 'cores' within the system. If you meet these criteria, please fill out this short survey, http://eee.uci.edu/survey/knx4e7LD5b, which will help OIT share information about HPC services, issues, and opportunities with cluster users and owners.