Python Acceleration: GPU-Powered Array Computing with CuPy

Accelerating array-based math by swapping NumPy for GPU-backed libraries.

In this tutorial, we will learn how to use CuPy to accelerate Python code with GPU computing. CuPy is a library that leverages NVIDIA CUDA to provide fast and efficient operations on large arrays.

We will cover the basics of CuPy’s features, how to perform array operations, and how to integrate CuPy with other libraries like NumPy and SciPy. By the end, you’ll understand how to use CuPy to significantly speed up your code’s performance on compatible hardware.

Prerequisites

1. Basic experience with Python (Programming with Python).

2. Basic experience with bash or similar Unix shells. (Software Carpentry - The Unix Shell )

3. User experience on Gadi. (Seminar on Getting Started with Gadi)

Learning Outcomes

At the completion of this training, you will be able to 

• applying CuPy features

• perform array operations

• integrating with libraries like NumPy

• understand how to significantly speed up your code’s performance on compatible hardware