Accelerated Python

Speed up your Python code with compilers and parallel execution.

Python has become the dominant language in scientific computing thanks to its high-level syntax, extensive ecosystem, and ease of use. However, its performance often lags behind traditional compiled languages like C, C++, and Fortran, as well as newer contenders like Julia and Chapel. This course is designed to help you speed up your Python workflows using a variety of tools and techniques.

We’ll begin with classic optimization methods such as NumPy-based vectorization, and explore just-in-time compilation using Numba, along with performance profiling techniques. From there, we’ll delve into parallelization – starting with multithreading using external libraries like NumExpr and Python 3.13’s new free-threading capabilities – but placing greater emphasis on multiprocessing.

Next, we’ll dive into Ray, a powerful and flexible framework for scaling Python applications. While Ray is widely used in AI, our focus will be on its core capabilities for distributed computing and data processing. You’ll learn how to parallelize CPU-bound numerical workflows – with and without reduction – as well as optimize I/O-bound tasks. We’ll also explore combining Ray with Numba and will discuss coding tightly coupled parallel problems.

Finally, we will learn how to use the Numba compiler to harness the full power of NVIDIA GPUs without leaving the Python ecosystem. We’ll cover high-level array acceleration, writing CUDA kernels, managing data transfer and basic performance tuning techniques. As a capstone exercise, we implement prime factorization on a GPU, with a focus on monitoring and optimizing GPU utilization on a cluster.

Date and time:
Friday, June 5, 2026, 9:30am-4:30pm (all day with a one-hour lunch break 12:30pm-1:30pm)

Location:
SFU’s Big Data Hub Presentation Studio (ASB 10900)

Instructor:
Alex Razoumov

Prerequisites:
Some knowledge of basic Python, numpy, and working with the Slurm scheduler.

Software:
We will provide access to one of our Linux systems. To make use of it, attendees will need a remote secure shell (SSH) client installed on their computer. On Windows we recommend the free Home Edition of MobaXterm. On Mac and Linux computers, SSH is usually pre-installed (try typing ssh in a terminal to make sure it is there).

Workshop material:
Towards high-performance Python
Distributed computing with Ray
Numba JIT compilation to CUDA kernels