pycuda-example

A small working CUDA example powered by Python and PyCUDA. Go to main.py to view the code.

The example implements an algorithm that counts the relatively non-prime numbers (GCD is not 1) for each number up to size=1000.

Also in the file main.cu a complete native CUDA code attached that does the same as in the python example.

Requirements

• NVIDIA device
• NVIDIA driver installed (https://www.nvidia.com/en-us/drivers/)
• CUDA installed (https://developer.nvidia.com/cuda-downloads)
• Python 3.9 (or higher)
• numpy>=2.0.2
• pycuda>=2024.1.2

For running C example only, first 3 points above will be enough.

How to run

Python example in main.py

1. Create a virtualenv: python -m virtualenv .venv
2. Activate the virtualenv: source ./.venv/bin/activate (or .\.venv\Scripts\activate for Windows)
3. Install requirements: pip install -r requirements.txt
4. Run the script: python main.py

C example in main.cu

1. Compile the code: nvcc -arch=native main.cu -o main (or nvcc main.cu -o main.exe for Windows)
2. Run the exe-file: ./main (or .\main.exe for Windows)