SMACK 12 - Do it faster! Simple ways to use all those cores (and GPUs) efficiently.
We will introduce several ways in which trivially embarrassingly parallel tasks can be run in laptops and desktops. We will introduce command-line tools such as GNU parallel and Kiko. We will then focus on simple techniques for optimisation of scientific computations in python. We will cover parallel computing with multiprocessing, acceleration of functions via numba, and GPU computing with cupy. The goal is to provide an easy roadmap for python code optimisation methods that can applied on already existing code, without writing a single line of C or FORTRAN.
About the talk