Sycamore Cluster

New as of January 2025, MPI capable CPU compute cluster that replaces Dogwood.

Sycamore is a Linux-based computing system available for Research at Carolina with

• more than 14,000 compute cores
• 117 TB system RAM across 78 nodes
• 400-gbps low latency high bandwidth network fabric
• fast /work file system

Sycamore provides an environment optimized for large, multi-node, tightly coupled computational models using a message passing (e.g. MPI) or hybrid (OpenMP + MPI) programming model.

Two different CPU types are available, Epyc 9654 and Epyc 9684x

Technical details can be found in the Sycamore System Overview (PDF).

Resources are managed through a fair-share algorithm using SLURM as the resource manager, or job scheduler.

Get an account on Sycamore

Getting started on Sycamore

Sycamore SLURM examples

Sycamore MPI modules

Sycamore Partitions and limits