System Specifications
Compute Nodes
ASPIRE 2A+ is built on an NVIDIA DGX SuperPOD™ system comprising 40 DGX H100 compute nodes. Each node has the following specifications:
| Component | # Nodes |
|---|---|
| CPU | 56 cores per CPU 112 core per node |
| GPU | 8x NVIDIA H100 80GB SXM |
| RAM | 2TB |
| Local Storage | 8x 3.84TB NVMe U.2 SED in RAID 0 array |
| Network | 8x NVIDIA ConnectX-7 single port InfiniBand cards Up to 400 Gb/s per card Up to 400 GB/s per node |
| NVSwitch | 4x 4th generation NVLinks 900 GB/s GPU-to-GPU bandwidth |
| Operating System | DGX OS Ubuntu 22.04 |
Network
NVIDIA Quantum 2 base NDR InfiniBand
Storage
Warning
All data on NSCC systems is governed by the Data Management and Retention Policy. Read it carefully to prevent data loss, maintain proper access, and manage your data securely.
| Directory |
Filesystem |
Capacity |
Mount Point |
Quota |
Use Case |
|---|---|---|---|---|---|
| Home | Lustre | 27.5 PB |
|
50 GB per user | Long term storage of user data |
| Project | /data/projects/<project-id> |
Per project allocation | Long term storage of project data shared among members | ||
| Scratch | Lustre | 2.5 PB |
|
100 TB per user |
Temporary storage of user data for better I/O performance (subject to 30-day purge) |
| Node Local NVMe | Ext4 | 28 TB per node |
|
- | Temporary storage of compute job data for better I/O performance (automatically removed at job end) |
Further Reading
To explore the architecture and components of ASPIRE 2A+ in greater detail, consult the following resources: