This workshop focuses on GPU accelerated computing with Python. Despite being a very popular language, Python not only is considered to be slow for high performance computing but also lacks GPU support. However, the use of Python-based libraries, such as CUDA Python (PyCUDA) and Numba, warrants further consideration on the subject of GPU computing with Python. This gives the user the best of both worlds: rapid development with Python combined with the speed of a compiled language targeting both CPUs and GPUs.

Upon completion, participants should understand key GPU concepts and be able to write simple GPU programs with Python.

Registration

• Academic participant : 10$
• Non-academic participant : $200

Prerequisites

A solid knowledge of Python basics is required. Please refer to the workshop PYT101 Programming with Python.

Course plan

• 1. Why computing on GPUs ?
• 2. Understanding the GPU architecture;
• 3. Very short intro to CUDA;
• 4. NumPy: library for arrays
• 4. Numba: just-in-time compiling;
• 5. Numba functions and compilation modes;
• 6. Numba+CUDA: your first GPU code;
• 7. Other options: CuPy, PyCUDA, CUDA C

Instructor

Nikolai Sergueev, analyst in advanced research computing at Calcul Québec.

Language

English

Technical prerequisites

We will use the Zoom platform. Because this event is a practical workshop, it is very useful having a secondary screen where you would get the instructor window on one screen and your own window on your main screen.

We will use the Jupyter Lab interface. Make sure you have a modern Web browser like Google Chrome, Firefox, Edge or Safari.

Contact

For any question, please write to training@calculquebec.ca.