Metis Cluster
Metis (commissioned in September 2023) is a 32-node CPU/GPU hybrid cluster running the Red Hat Enterprise Linux 8.x operating system. Each compute node is an HPE DL 385 Gen 10+ V2 server equipped with:
- 2x AMD EPYC 7713 CPUs 2.0 GHz 64-core processors
- 1259 GB RAM, 1 x 4TB SSD scratch disk drives
- 1 x NVIDIA A100 GPU, Amper™ architecture, 40 GB RAM each card
- All 32 nodes are connected via a 200 Gbps Infiniband network
A Cray ClusterStor E1000 storage server provides the cluster with 1 PB of shared disk space.
Detailed specifications of the system can be found here:
Gaea Cluster
Gaea is a 60-node CPU/GPU hybrid cluster running the Alma Linux 8.x operating system. Each compute node is an HP SL380s G7 equipped with:
- 2 x Intel X5650 2.66 GHz 6-core processors
- 144 GB RAM, 4 x 500 GB 2.5" SATA disk drives in RAID10 configuration (i.e., 1 TB each node)
- 2 x NVIDIA TESLA P4 GPU, Pascal™ architecture, 8 GB RAM each card
- All 60 nodes are connected via Full 1:1 non-blocking Infiniband and Ethernet switch connectors.
The cluster also has two HP Proliant DL380G7 storage servers and an HP P2000 disk storage array with 192 TB of effective storage space (i.e., after RAID6). The storage array connects the servers via 6 Gigabit per second SAS connections.
NICADD Cluster
NICADD is a 700 processor slots (1.8-2.6 GHz) cluster under the HT CONDOR batch system running Alma Linux 8.x OS. This facility also provides three interactive nodes with instant access to CERN, FERMILAB and OSG software libraries. The cluster is operated by the Northen Illinois Center for Accelerator and Detector Development. For access and documentation please consult with pages at nicadd.niu.edu/nhpc.