High-performance computing (HPC) upgrade complete to mark 10-year anniversary
UCT’s new HPC cluster was installed and operational in January 2019 and continues to provide a reliable, scalable and economic computing facility for researchers. The upgraded facility provides an HPC cluster geared towards researchers with major computing needs but is also accessible to the full community, for both teaching and research.
“June 2019 marked ten years since we started providing HPC services to UCT researchers,” says Andrew Lewis, a senior technical specialist at ICTS. “In those 10 years we’ve run almost 26 million CPU hours, completed just under 3 million jobs and have been acknowledged in 138 publications that we’re aware of.”
The research infrastructure landscape has changed considerably since 2009, says Timothy Carr, a senior technical specialist at ICTS. “We no longer have to rely on hand-me-down servers and cobbled-together file services. Our new cluster has been running for six months now and has seen a very good uptake of new researchers.”
Given the cost of electricity both for powering and cooling central processing units (CPUs) this affordable and environmentally-friendly computing resource is of great value to researchers.
The new cluster is split into several partitions, featuring newer, faster nodes; graphic processing units; high-memory nodes; and the older, original cluster nodes now available for teaching and lower-priority jobs. It provides state-of-the-art 100GB Infiniband interconnect for both storage and inter-process communication as well as increased storage space for temporary processing.
The HPC team provide the following support for UCT researchers:
- Account creation on UCT Linux HPC infrastructure.
- Basic introductory training to get users started.
- Access to basic Linux skills and advanced HPC training.
- Installation of scientific codes (software applications).
- Access to fast IO scratch disk for processing of data.
- Access to archive storage.