Deepthought: System Specifications
SLURM Partition Layout
Important
The partition list below represents the current hardware that we have that will be available once the remaining hardware has been relocated from the existing cluster
| Description | Partition Name | Nodes | Default Run Time | Max Run Time |
|---|---|---|---|---|
| General Compute (CPU) | general | 12 | 60 minutes (0-01:00:00) | 7 days (7-00:00:00) |
| CUDA Compute (GPU) | gpu-cuda | 2 | 60 minutes (0-01:00:00) | 14 days (14-00:00:00) |
| High Capacity (CPU) | high-capacity | 5 | 60 minutes (0-01:00:00) | 14 days (14-00:00:00) |
| IDS System Support | hpc-support | all | Unlimited | Unlimited |
| Michelle Coote Lab | mcoote | 5 | 60 minutes (0-01:00:00) | 14 days (14-00:00:00) |
| Open OnDemand Interactive Apps | ood | 4 | 60 minutes (0-01:00:00) | 12 hours (0-12:00:00) |
| Short Runtime Jobs | short | 4 | 60 minutes (0-01:00:00) | 1 day (1-0:00:00) |
| Teaching | teaching | 4 | 60 minutes (0-01:00:00) | 12 hours (0-12:00:00) |
Storage Layout
Home Drives (NFS): ~22TB mounted on all nodes
Modules, Software and System Files (NFS): 3TB mounted on all nodes
Semi-persistent Dataset Storage (Ceph): ~150TB mounted on all nodes
Temporary Scratch/Job Storage (BeeGFS): ~250TB mounted on all nodes
Note
Network Drives (R/T) are available on the login node but not on individual compute nodes
Node Breakdown
1x Login Node with 32 Cores / 64 Threads and 384GB of RAM
16x General Compute Nodes each with 64 Cores / 128 Threads and 768GB of RAM
5x High Capacity Nodes each with 64 Cores / 128 Threads and 3TB of RAM
4x NVIDIA L40S GPUs each with 48GB of RAM located in pairs in two of the High Capacity Nodes