NGH Cluster
CCI has obtained two GraceHopper nodes for intial beta testing.
These will be made available to select users to assist with our testing.
Limitations¶
- No shared filesystem and limited local storage. This means you will not have access to your home folder or existing software install.
- No scheduler. You will be assigned a node for a limited time period.
- CCI modules are not available. You will need to install your own software.
- OS on these nodes is currently Ubuntu 22. Long term they will switch to RHEL like the rest of CCI.
System information¶
System contains 2 nodes, each with:
- 1x Nvidia Arm GraceHopper CPU/GPU Superchip
- 1x Grace Arm v9 CPU with 72 cores
- 1x H100 Hopper GPU with 96GB HBM3e
- 480GB LPDDR5X ECC RAM
- 1x 100Gb Ethernet
- NVLink-C2C: 900GB/s coherent memory
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.154.05 Driver Version: 535.154.05 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 NVIDIA GH200 480GB On | 00000009:01:00.0 Off | 0 |
| N/A 26C P0 64W / 900W | 1MiB / 97871MiB | 0% Default |
| | | Disabled |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| No running processes found |
+---------------------------------------------------------------------------------------+