• Connecting to an HPC
  • Account management
    •  
    • environment ( modules )
    • account usage, project participants, quotas
  • Programming environment
    • traditional HPC languages (scale well, high efficiency): c and fortran
    • python, machine-learning
    • commercial software
  • Data Transfer with an HPC
    • globus online
    • ssh/scp/rdist
  • Thinking about your workflow end to end
    • optimize performance 
      • on the HPC
        • performance profiling
      • interacting with the HPC
        •  
        • data transfer bottlenecks (last mile )
          • WiFi
          • USB-N with N < 3 ?
        • on a campus lan
          • gigabit ethernet ( show bandwidth and latency from a test )
        • on a "foreign" lan (home, coffee shop, another continent...)
          • vpn ?
            •  
            • performance vs security
    • debugging codes and issues that arise

References:

https://bluewaters.ncsa.illinois.edu/user-guide

https://campuscluster.illinois.edu/resources/docs/user-guide/


  • No labels