As of April 2023, the RIT-HPC Compute Cluster is comprised of:
Software
- Base operating platform: Centos 7 (currently upgrading to AlmaLinux 8)
- All compute nodes are virtual machines running under Openstack.
- Scheduler: Slurm
Compute
- 21,000 hyperthreaded cores- Xeon E5-2670 v2 @ 2.50GHz and Xeon Gold 6140 CPU @ 2.30GHz
- 460 compute nodes
- Compute nodes typically have 128GB, 192GB or 256GB of RAM.
- Low number of large compute nodes, typically with 512GB or 768GB of RAM
- 8 x GPU Nodes: 1 x NVIDIA V100 32GB Telsa GPU, 19 cores, 180GB RAM, 1.2TB local scratch space
Network
- Interconnects: multi-bonded 10GE and 40GE
- Core: HPE 12908 IRF
- Top of Rack: Mellanox sw6036 & Dell 4000 series
Storage
- Tier 1 storage – Isilon Gen6 platform, H500 & A2000 storage nodes, 14PiB
- Cluster access to Tier 1 storage: NFSv3
- End user access to Tier 1 storage: SMB
- Archive storage – iRODS. 12PB usable with offsite replica