¶ System Overview
Our High-Performance Computing (HPC) cluster provides substantial computational resources for research and analysis. The cluster consists of multiple components working together to provide a powerful computing environment.
¶ Cluster Components
- Login Node
- Entry point for all users
- Used for:
- Job submission
- File management
- ⚠️ Not for running intensive computations
- Compute Resources
- CPU Nodes (Glacier)
- 38 dedicated CPU compute nodes
- Used for CPU-intensive calculations
- Ideal for parallel processing tasks
- GPU Nodes
- Dogfish Nodes: 2 GPU NVIDIA A100
- Puffin Nodes: 2 GPU NVIDUA A30
- Specialized for machine learning and GPU-accelerated computations
- CPU Nodes (Glacier)
- Job Scheduler
- SLURM Workload Manager - Read more here
- Manages resource allocation
- Handles job queuing and execution
¶ Cluster Architecture
¶ Cluster File System Structure
¶ ⚠️ Important Storage Warning
Your home directory (~) has a strict 5GB limit! If it gets full:
- You may be unable to log in
- Jobs might fail to submit or run
- Shell configurations might not load
- Software may crash or behave unexpectedly
¶ Directory Organization
/home/username/ # or ~ (5GB LIMIT - USE CAREFULLY!)
├── .zshrc # Shell configuration files
├── .bashrc # (if using bash)
└── essential_files/
/sci/labs/ # Fast storage without backup
└── PI-USERNAME/ # Lab directory (1TB)
├── lab_member1/
├── lab_member2/
├── lab_member3/
└── shared/ # Shared lab resources¶ Directory Details
¶ Home Directory (~ or /home/username/) - 5GB LIMIT!
- ⚠️ NOT recommended for computational work
- Use only for:
- Configuration files (
.zshrc,.bashrc, etc.) - Essential scripts
- Small configuration files
- Configuration files (
¶ Lab Storage (/sci/labs/PI-USERNAME/)
- Fast storage space without backup
- 1TB allocation per lab by default (More quota options read here)
- Recommended for all computational work
- Structure:
/sci/labs/PI-USERNAME/
├── lab_member1/ # Personal workspace (USE THIS!)
│ ├── projects/ # Active projects
│ ├── data/ # Working data
│ └── results/ # Analysis outputs
├── lab_member2/
├── lab_member3/
└── shared/ # Shared lab resources
├── datasets/ # Common datasets
├── software/ # Lab-specific software
└── results/ # Shared results