Thank you for your interest in contributing to the CUDA Kernels project! We welcome improvements, bug fixes, and new learning modules.
- Fork the Repository: Create your own copy of the project.
- Create a Branch: Use a descriptive name for your feature or fix (e.g.,
feature/matrix-mult). - Implement Your Changes:
- Ensure your code follows the existing modular structure.
- Use the shared
CHECKmacro for error handling. - Provide a local
README.mdif introducing a new module.
- Verify: Build your module using the root
Makefile. - Submit a Pull Request: Describe your changes and explain their benefits.
- Keep kernels focused on a single concept.
- Document parallel algorithms clearly within the code.
- Ensure cross-platform compatibility where possible (Windows/Linux).