Loading Events

Online Training Webinar: Python for HPC

12 April 2019
9:30am - 11:30am
Event cost: Free

Registrations have now closed!

Description

It’s no surprise that Python is one of the most popular languages amongst scientists; it’s an easy language to learn, has a rich software ecosystem, and can provide relatively good performance.  However, researchers often run into issues when using Python in an HPC setting.  This webinar will focus on how to help researchers effectively use Python in their HPC workflows at Pawsey.

We’ll discuss some of the good (and bad) things Python does, cover different Python modules that researchers should be using to get good performance, and also discuss different ways of writing parallel Python code.

The course will mostly be run via Jupyter notebooks, allowing participants to work hands-on for most of the webinar.

To simplify software dependency issues, Pawsey staff will be providing a Docker container with the required Python modules and Jupyter notebook server.  If you want to use this container, you’ll need to have Docker installed on your laptop (https://www.docker.com/get-started).  This container will also be used to run examples on Pawsey’s Zeus system.

You are welcome to use your own Python modules if you wish (e.g. in Anaconda), but you’ll need to install the required packages before-hand (a complete list will be made available shortly).

Agenda

  1. Overview of Python and how it fits into an HPC workflow (e.g. memory management, data structures, compiled vs. interpreted code)
  2. Hands-on examples (via Jupyter notebooks) in the following topics:
    • Computational modules (NumPy, SciPy, Scikit)
    • I/O modules (pytable, h5py)
    • Parallel processing tools (Multiprocessing, MPI4PY, Numba, Cython)

Please note registrations will close at 4pm AWST on Wednesday 10th April.

Please complete the registration details on the form below.

Registrations should be made with institutional/business email addresses.