Fundamentals of Accelerated Computing with CUDA Python

This repository contains materials and exercises for the "Fundamentals of Accelerated Computing with CUDA Python" course.

Introduction

This course provides hands-on experience accelerating Python applications using NVIDIA CUDA. You'll learn how to leverage NVIDIA GPUs to significantly improve the performance of your Python code through parallel computing techniques.

Prerequisites

• Basic Python programming knowledge
• Familiarity with NumPy
• Understanding of basic parallel programming concepts (helpful but not required)
• NVIDIA GPU with CUDA support
• CUDA toolkit installed

Repository Structure

This repository contains:

• Jupyter notebooks with guided exercises
• Sample solutions
• Supporting Python modules and utility files
• Data files used in exercises

Setup Instructions

1. Install the required dependencies:

   pip install numba cuda-python cupy
   

2. Verify your CUDA installation:

   python -c "import numba; print(numba.cuda.is_available())"
   

3. Launch Jupyter notebooks:

   jupyter notebook
   

How to Use

Work through the notebooks in sequential order. Each notebook introduces new concepts and builds upon previous lessons. Complete the exercises in each notebook to reinforce your learning.

Course Completion

Upon completing all exercises in this repository, you will have gained practical experience in:

• Profiling Python applications to identify bottlenecks
• Using Numba to accelerate Python functions with CUDA
• Working with CUDA memory management
• Implementing parallel algorithms on the GPU
• Optimizing CUDA Python code for maximum performance

Attribution

These materials are based on the NVIDIA Deep Learning Institute (DLI) course: "Fundamentals of Accelerated Computing with CUDA Python".

Original course can be found at: NVIDIA DLI - Fundamentals of Accelerated Computing with CUDA Python

Copyright © NVIDIA Corporation. All rights reserved.