| System Name | Description | Storage | Networking | Hostname | Status | Service dates | Access |
|---|
| Clusters |
| HAL | Cluster of 16 IBM POWER9 nodes with 4 NVIDIA V100 GPUs per node | 244 TB usable, NVME SSD-based storage by DDN Peak system bandwidth ~100GB/s | 10GbE external Dedicated 10GbE for data transfer dual-channel EDR IB internal | hal | in service | 2019 | open |
| KNC | Xeon Phi 5-node co-processor cluster |
|
| knc | not in service |
|
|
| KNL | Xeon Phi 4-node processor cluster (access via KNC) |
|
| knl* | not in service |
|
|
| ISL Cluster | various x86 resources under a single login and scheduler | 122 TB usable | 10GBe external EDR IB internal | isl-login01 | in beta | 2024- | open, but allocated |
| Clouds |
| VLAD OpenStack cluster | OpenStack Cloud deployment |
|
|
| in service |
|
|
|
|
|
|
|
|
|
|
| Servers |
| DGX A100 | NVIDIA DGX A100 | 14TB RAID0 | 1GbE | hal-dgx | in service | 2021- | open |
| Ac33 | TYAN server with Intel Westmere-EP CPUs and NVIDIA Tesla C2050 GPUs | Local 300GB + 1TB HDD |
| ac33 | not in service | 2014 | open |
| Hybrid | SuperMicro server with Intel Haswell-EP CPUs and NVIDIA Tesla P100/V100 GPUs | Local 75GB SSD + one (3x 4TB) RAID 0 arrays | 10GbE | hybrid | in service | 2016 | restricted |
| Vlad-hm | High memory server with 3TB of ECC RAM | Local 3TB SSD + two (6x 1TB) RAID 0 arrays CephFS 257TB (reported) | 10GbE | vlad-hm | not in service
|
| deprecated (use isl-cluster) |
| AMD ARM server |
|
|
| not in service |
|
|
| Power9 AC922 server | IBM POWER9 4x NVIDIA Tesla V100 16GB | Local 2x 3.84TB SSD Local 1.4TB NVME | 1GbE | p9 | in service |
| restricted to XPACC project |
NVIDIA ARM HPC DevKit (overdrive) | 1x Ampere Altra Q80-30 2x NVIDIA A100 GPU |
|
|
|
|
| restricted to NCSA users only |
| FPGA dev |
| Altera DE5 FPGA board |
|
|
|
|
|
|
| pynq | XILINX PYNQ-Z1 |
|
| pynq | offline | 05/2017 | restricted |
| XilinxDevKit | Xilinx Kintex UltraScale FPGA KCU1500 Acceleration Development Kit |
|
| iridium | in service |
| restricted |
| CAPI2.0 platform | Nallatech 250S+in Power9 server (hal000) 4x NVIDIA Tesla V100 16GB | Local 2x 3.84TB SSD | 10GbE | hal000 | in service | 2018 | restricted to NSF MRI project |
| OpenCAPI platform (IBM P9 IC922) |
|
|
|
|
|
|
|
| Xilinx u250 | x86 server with Alveo U250 Data Center Accelerator Card and Vitis development environment |
|
| hal-fpga-x86 | in service | 2020 | restricted to NSF MRI project |
| Misc |
| Ceph storage |
|
|
| offline |
|
|
| Raspberry Pi boards |
|
|
| offline |
|
|
| Jetson | NVIDIA Jetson TK1 cluster |
|
|
| offline |
| restricted |
| Jetson TX2 | NVIDIA Jetson TX2 cluster | 1TB NFS-mounted /home | 1GbE internal & external | jetson-tx2-* | offline |
| open |
|
|
|
|
|
|
|
|