General User guides

• IDSC Cluster Overview
  • Triton Supercomputer
  • Pegasus Supercomputer
• Important Guidelines
  • Review Policies and Project Requirements
  • Network Access Requirement
  • Job Submissions and Scheduler Guidance
  • Software Modules and Environment Management
  • File Transfer Best Practices
  • Scratch Usage for Job I/O
• Connection
• Projects & Resources
• Job Scheduling
  • LSF Overview
  • Queues
    • Pegasus Queues
    • Triton Queues
  • Commands
  • Submitting Jobs
  • Monitoring Jobs
  • Examining Jobs
  • Interactive Jobs
  • Serial/Parallel Job Scripts
  • How to Request GPUs
  • Memory Management
    • Cluster Hardware and Queue Model
      • Standard compute nodes
      • Big-memory nodes (bigmem)
      • Queues and host spanning rules
    • Units and what is enforced
      • Memory units
      • Reservation vs. limit
    • How rusage[mem=…] behaves on Pegasus
      • Per-host reservation (the most important rule)
    • Controlling placement (single host vs multiple hosts)
      • Single-host placement
      • Multi-host placement (parallel jobs)
    • A simple formula to apply (parallel queue)
    • Guidance by queue
      • Queue: normal / general (single host only)
        • CPU guidance
        • Memory guidance
        • Examples
      • Queue: parallel (multi-host allowed)
        • Examples
      • Queue: bigmem (96 cores / 2 TB RAM)
        • Single-host bigmem jobs
        • Multi-host bigmem jobs
    • Verifying what you asked for and what happened
      • Show what resources were requested
      • Confirm placement and packing
      • Diagnose pending jobs
    • Recommended starting points (Pegasus)
      • Standard node jobs (single host)
      • Parallel jobs
      • Bigmem jobs
    • Summary
• Quickstart Guide
  • Basic Concepts
    • Home Directory vs Scratch Directory
    • Login Node vs. Compute Node
  • Basic Workflow
    • 1. Log in to the Pegasus and Triton Systems using SSH
    • 2. Prepare Your Code
    • 3. Preparing the input data
    • 4. Create a Conda Environment
    • 5. Prepare the Job Script
    • 6. How to submit job to LSF and Monitor LSF job
    • 6. Checking the job output
    • 7. Logging out from Triton or Pegasus
• Parallel Computing
  • Loading the Appropriate MPI Module
  • Sample parallel programs:
    • C++ source code and compilation
    • C source code and compilation:
    • Fortran 90 source code and compilation:
    • Fortran 77 source code and compilation:
  • The LSF script to run parallel jobs