Release the GIL! Parallel Geometric Constraint Solving with Cython

2:25pm - 2:55pm on Friday, October 2 in Online

Maryanne Wachter

Description

One of the common complaints about Python is that while it is approachable, it is impossible to match the performance of a compiled language such as C or C++. However, using Cython allows you to have the best of both worlds. You can preserve the structure and readability of a Python package with the performance of a C/C++ program. Common pitfalls of Cython will be discussed through the lense of adapting an existing geometric constraint solving C++ library used widely in both architectural design and structural engineering. During this talk, particular emphasis will be placed on how this project was structured to allow for releasing the GIL to parallelize specific solve methods for a 100x speedup in simulation runtimes compared to pure Python.

The goal of this talk is to show how Cython can be used for implementing powerful algorithms in Python to develop code that is approachable and extensible.

Want to edit this page?