Professional Native CUDA + OpenMP Template

A high-performance, modern template for local GPU development using CUDA and OpenMP. Optimized for single-node performance with a professional build system and developer-friendly utilities.

Features

• Modern CMake: Target-based configuration with C++17, CUDA 11+, and OpenMP support.
• Hybrid Computing: Utilize both GPU (CUDA) and CPU (OpenMP) parallelism on a single node.
• Environment Configuration: Type-safe environment variable loading via .env files.
• Optimized Memory: Custom PinnedAllocator for fast Host-to-Device (H2D) and Device-to-Host (D2H) transfers.
• Cross-Platform Makefile: Simple, standardized commands for building, running, and formatting.
• Utility Suite: ANSI colors, CUDA error checking macros, and environment parsing.

Directory Structure

• src/: Source files (.cpp, .cu).
• include/: Public headers.
• utils/: Internal utility headers (allocators, colors, etc.).
• examples/: Main application entry points and demonstrations.
• lib/: Third-party libraries (submodules).

Getting Started

1. Configure

Create your local environment file:

cp .env.example .env

2. Build

make build

3. Run

make run

4. Format Code

Maintain consistent styling using clang-format:

make format

Environment Variables (.env)

Customization

• Project Name: Edit project(CUDA_Project ...) in CMakeLists.txt.
• Target Architecture: The build system uses the -real suffix (e.g., 75-real) to ensure compatibility and performance.
• GPU Monitoring: Use make gpu-status to monitor usage via nvidia-smi.

Author

nguyenpanda
Email: [email protected]
Website: https://nguyenpanda.com

License

MIT License