System Hardware
Category | Specifications |
Computer Nodes | 443 |
GPU Nodes | 27 |
Login Nodes | 2 |
Large Memory Nodes (>512GB) | 4 |
Total CPU Cores | 13100 |
Total GPUs | 199 |
Total Storage (GPFS Filesystem) | 3.27PB |
Operating Systems | RedHat EL 7.3 (Linux) |
Job Scheduler | Slurm Workload Manager |
Oscar has 425 compute nodes including 27 GPU nodes in the batch, gpu, gpu-he, vnc, bigmem, and debug partitions.
Nodes could be down for repair. Use the sinfo command to check node status on Oscar.
The batch partition is for programs/jobs which need neither GPUs nor large memory. There are 362 nodes in the batch partition.
Model | Cores per Node | Memory (GB) | Node Count | GPU (GPU Memory) | Nodes |
Intel Broadwell (ES-2650 v3) | 20 | 125 | 4 | n/a | node906, node910, node925-926 |
Intel Broadwell (ES-2650 v4) | 24 | 125 | 51 | n/a | node1003-node1053 |
Intel Skylake (Gold 6126) | 24 | 187 | 44 | n/a | node1101 - node1120, node1122 - node1128, node1161 - node 1164, node1314, node1317-node1328 |
Intel Skylake (Gold 6126) | 24 | 93 | 47 | n/a | node1121, node1129 - node1160, node1301 - node1313, node1315 |
Intel Cascade Lake (Gold 6242) | 32 | 376 | 40 | n/a | node1333 - node1364, node1601 - node1608 |
Intel Cascade Lake (Gold 6242) | 32 | 187 | 100 | n/a | node1613-1656, node1701-1756 |
Intel Cascade Lake (Platinum 8268) | 48 | 376 | 76 | n/a | node1801-1864, node1901-1912 |
Total | 10960 | ​ | 332 | ​ | ​ |
The bigmem partition is for programs/jobs which require large memory. There are 4 nodes in the bigmem partition.
Model | Cores per Node | Memory (GB) | Node Count | GPU (GPU Memory) | Nodes |
Intel Cascade Lake (Gold 6242) | 32 | 2096 | 2 | n/a | node1609-node1610 |
Intel Cascade Lake (Gold 6242) | 32 | 754 | 2 | n/a | node1611-node1612 |
Total | 128 | ​ | 4 | ​ | ​ |
The debug partition is for users to debug programs/jobs. There are 7 nodes in the debug partition.
Model | Cores per Node | Memory (GB) | Node Count | GPU (GPU Memory) | Nodes |
Intel Haswell (E5-2650 v3) | 24 | 125 | 7 | n/a | node970-node976 |
Total | 168 | ​ | 7 | ​ | ​ |
The gpu partition is for programs/jobs which require GPUs. There are 19 nodes (145 GPUs) in the gpu partition.
Model | Cores per Node | Usable Memory (GB) | Node Count | GPUs Per Node | GPU (GPU Memory) | Nodes | X Server |
Intel Skylake (Gold 5122) | 8 | 93 | 4 | 4 | titanv (12G) | gpu1201 - gpu1204 | Y |
Intel Skylake (Gold 5122) | 8 | 187 | 1 | 4 | p100 (12G) | gpu1209 | Y |
Intel Cascade Lake (Gold 6142) | 32 | 376 | 2 | 5 | titanrtx (24G) | gpu1401 - gpu1402 | Y |
Intel Skylake (Gold 6126) | 24 | 187 | 1 | 5 | p100 (12G) | gpu1403 | Y |
Intel Cascade (Platinum 8268) | 48 | 376 | 1 | 10 | quadrortx (24G) | gpu1414 | Y |
Intel Cascade Lake (Gold 6242) | 32 | 376 | 10 | 10 | quadrortx (24G) | gpu2001 - gpu2010 | Y |
Total | 496 | ​ | 19 | ​ | ​ | ​ | ​ |
The gpu-debug partition is for users to debug gpu programs/jobs. There is 1 node (4 GPUs) in the gpu-debug partition.
Model | Cores per Node | Memory (GB) | Node Count | GPUs Per Node | GPU (GPU Memory) | Nodes | X Server |
Intel Skylake (Gold 5122) | 8 | 187 | 1 | 4 | p100 (12G) | gpu1207 | Y |
​ | 32 | ​ | 1 | ​ | ​ | ​ | ​ |
The gpu-he partition is for programs/jobs which need to access high-end GPUs. There are 5 nodes (28 GPUs) in the gpu-he partition.
Model | Cores per Node | Memory (GB) | Node Count | GPUs Per Node | GPU (GPU Memory) | Nodes | X Server |
Intel Skylake (Gold 6126) | 24 | 187 | 3 | 4 | v100 (32G) | gpu1210-gpu1212 | Y |
Intel Broadwell (E5-2698 v4) | 40 | 503 | 2 | 8 | v100 (32G) | gpu1404-gpu1405 | Y |
Total | 152 | ​ | 5 | 28 | ​ | ​ | ​ |
The vnc partition is for users to run programs/jobs in an graphical desktop environment. There are 27 nodes (16 GPUs) in the vnc partition.
Model | Cores per Node | Memory (GB) | Node Count | GPUs Per Node | GPU (GPU Memory) | Nodes | X Server |
Inet Broadwell (E5-2697A v4) | 32 | 755 | 2 | 8 | 1080ti (11G) | gpu717-718 | Y |
Intel Haswell (ES-2650 v4) | 24 | 125 | 25 | n/a | n/a | node945-node969 | N |
Total | 664 | ​ | 27 | 16 | ​ | ​ | ​ |
