AiMOS Quick Start Information

The IBM Research AI Hardware Center is a global research hub headquartered in Albany, New York. The center is focused on enabling next-generation chips and systems that support the tremendous processing power and unprecedented speed that AI requires to realize its full potential. AiMOS (short for Artificial Intelligence Multiprocessing Optimized System) serves as the test bed for the IBM Research AI Hardware Center.

Contents:

• AiMOS Overview
  • What are the landing pad nodes?
  • What are the front end nodes?
  • What are the compute nodes?
• Access to AiMOS
  • Apply for a user ID on AiMOS
  • Login to AiMOS
• AiMOS Environment Basics
  • Initial Environment Setup
  • The GPFS shared file systems
  • Slurm Resource Manager and Job Scheduler in AiMOS
• Set up the workload environment
  • Load additional tools and libraries via module
  • Install Anaconda
  • Frequently Used conda commands
  • Create a new conda environment for your workload
  • Install WML-CE (a.k.a PowerAI)
  • Install Open-CE
  • Install Jupyter notebook
  • Start Jupyter Notebook
  • Display the WebGUI for a Jupyter notebook via Tunneling
  • Remote Port Forwarding via SSH Tunneling
• Submit jobs
  • Prerequisites
  • Interactive Job
  • Batch job
  • Build the job dependencies
  • Request for up to 48 hours run time
  • Request for NVMe storage on the compute nodes
• AI Benchmarks
  • Common AI Datasets (CADS) Repository
  • Common AI Benchmarks (CABS) Repository
  • Monitoring Tools
• IBM Research AI Hardware Toolkit
  • Overview Analog AI Hardware Toolkit(AIHWKIT)
  • Training Analog Model
  • MNIST Training Example
  • More Examples
• Hints and Tips
  • General
  • DCS Cluster (Power)
  • NPL Cluster (X86)
  • Transfer data in and out of AiMOS
  • Conda Init in .bashrc
  • Sample .condarc
  • Troubleshooting Tips
• Getting Help?