High Performance Computing

High-performance computing (HPC) refers to large jobs with more or less tightly coupled subprocesses that can run in parallel and communicate with each other by sharing memory locations or by passing messages across a specialized network. Computing clusters are collections of processors (often mounted in racks) coupled together by a shared-memory buss or a high-speed network.

Clemson is in the process of developing several cluster resources:

• CCIT has pooled resources with several researchers on campus to purchase a large HPC cluster. This is known as palmetto or "The Condo Cluster." The guide shown in the menu to your left describes the cluster along with how to access and use it.
• CCIT manages and/or houses independent clusters for several researchers on campus. In return, spare cycles on these machines are available for use by the Clemson community.
  • axiom is the Mathematical Sciences Department's IBM OpenPower cluster.
    • 48 compute nodes, 4 interactive nodes, and a head node, each with dual-core Power5 64-bit processors, 4GB of memory and 80GB of disk
    • 550GB of shared RAID10 disk
    • Red Hat Enterprise Linux 5 Advanced Server, GNU and IBM XL compilers, torque and maui queue manager
    • Documentation
  • wcnitc is an Electrical and Computer Engineering Intel cluster.
    • 70 compute nodes, each with 4 Intel x86_64, 12GB of memory
    • 4TB shared disk
    • Red Hat Enterprise Linux 4 Advanced Server, GNU compilers, Condor
  • arev is the CU-CCMS Sun cluster.
    • 430 SunBlade x6250 nodes
    • Infiniband
    • SLES 10 SP1 64bit
  • aris is one of the CU-CCMS Sun Fire E6900 servers.
    • 24 UltraSparcIV+ processors
    • 384 GB shared memory
    • Solaris 10

Information about access to these resources is available here.