Getting Up and Running on MeluXina Supercomputer

This training covers HPC systems starting from architecture and system’s basic components to software stack using job schedulers and batch systems. Attendees will have the opportunity to get first hands-on experience on the MeluXina supercomputer. The morning session offers a refresher on the usage of Linux and computer architectures. Attendees comfortable with these topics can skip the morning session and directly attend the afternoon one.

At the end of this course, the successful attendee will:

• Have knowledge about:
  • Linux command line usage and bash script fundamentals
  • Basic computer architecture
  • The architecture of the MeluXina HPC system
  • The module system of MeluXina to manage the software environment
  • Basic usage of the SLURM job scheduler of MeluXina
  • Singularity containers and the benefit of using containers
• and be able to:
  • Connect to and send computational jobs to MeluXina
  • Build a parallel shared memory program and run it on MeluXina
  • Build a parallel message-passing program and run it on MeluXina
  • Transfer data to and from MeluXina
  • Monitor running jobs on MeluXina
• in order to:
  • Make effective use of the MeluXina supercomputer for computational problems

• Refresher on Linux and computer architecture
  • Introduction to Linux system
    • Introduction to command-line (Unix/Linux systems)
    • Bash script Fundamentals (variables/loops)
  • Basics of computer architecture
    • Going back to the basics. Some familiarity with notions such as CPU or memory helps to understand the specificities of a supercomputer such as MeluXina
• HPC architecture and usage
  • Diving in: Architecture of supercomputers
    • Components of a HPC system, Properties of Login node, Compute node, Storage node, HPC networks
    • Detailed, advantages and usability of NUMA node
    • Discussion about advantages of HPC in the current context
  • HPC: MeluXina system
    • Introduction to the module system (LMOD) to manage software environment on MeluXina
    • Introduction to the usage of SLURM job scheduler
    • Data storage spaces
  • Hands-on session, using MeluXina
    • Connection through command-line
    • Transferring data to MeluXina
    • Compiling
    • Submitting jobs (sequential/parallel/interactive)
    • Monitoring jobs
  • Singularity containers
    • Building a container
    • Transfer container to MeluXina
    • Running a container on MeluXina