Service Description
Research Computing supports the Longleaf and Dogwood compute clusters. The Longleaf compute cluster is designed for GPU, memory, and I/O intensive or loosely coupled workloads with an emphasis on aggregate job throughput over individual job performance. Workloads consisting of a large quantity of jobs, each requiring a single compute host, are best suited to Longleaf.
The Dogwood compute cluster is designed for high-speed interconnect-intensive, tightly coupled parallel workloads, such as those using message passing between nodes, e.g., with MPI. Workloads where a single job requires multiple compute hosts are best suited to Dogwood (parallel work).
Audience
Employees and Students
Expected Delivery
5 Business Days
Keywords
Sycamore, throughput, HPC, HTC