Researchers requiring large amounts of computing power to accomplish their science goals should target XSEDE's HPC resources. Codes that run in this category are tightly coupled MPI, OpenMP, GPGPU, and hybrid programs. These codes require many low latency interconnected nodes (massively parallel processing), large memory distributed across many compute nodes, large memory shared by multiple tasks (shared memory processing), and/or high-performance parallel file systems to accommodate simulation results. The aggregate simulation time ranges from a few hours to days.

Available HPC resources

PSC's Blacklight Blacklight
User Guide
Blacklight is a SGI shared memory system intended for applications that require a large shared memory for computational tasks.
SDSC's Comet Comet New!!
User Guide
SDSC's large Comet cluster is 100% dedicated to XSEDE users. Comet features the next generation Intel "Haswell" processors with AVX2.

Comet is now accepting requests for Trial Allocations.
NICS' Darter Darter
User Guide
Darter is a Cray XC30 system providing both high scalability and sustained performance with a peak performance of 250 Tflops.
Gordon Gordon
User Guide
Gordon is a unique, a flash-based supercomputer designed for data-intensive applications.
IU's Mason Mason
User Guide
Mason at Indiana University is a large memory computer cluster configured to support data-intensive, high-performance computing tasks using genome assembly software.
TACC's Stampede Stampede
User Guide
Debuting Intel's new innovative MIC technology on a massive scale, TACC's Stampede system commenced production in January, 2013.
LSU's SuperMIC SuperMIC
User Guide
Equipped with Intel's Xeon Phi technology, LSU's SuperMIC cluster consists of 380 compute nodes.
SDSC's Trestles Trestles offline April 30, 2015
User Guide
Trestles employs flash-based memory and is designed for modest-scale research providing very fast turnaround time. It is intended for moderately scalable parallel applications with an emphasis on improving productivity for a broad spectrum of users. Trestles is ideal for applications with fast local I/O requirements that can benefit from the flash memory available on each compute node.
TACC's Wrangler Wrangler Coming Soon
User Guide
The Wrangler Data Analytics System combines database services, flash storage and long-term replicated storage, and an analytics server to meet the needs of today's data scientists.

Last update: April 16, 2015