cuDNN FrontEnd(FE)

cuDNN FE is the modern, open-source entry point to the NVIDIA cuDNN library and high performance open-source kernels. It provides a C++ header-only library and a Python interface to access the powerful cuDNN Graph API and open-source kernels.

🚀 Embracing Open Source

We will begin open-sourcing kernels based on customer needs, with the goal to educate developers and enable them to customize as needed.

We are now shipping OSS kernels, allowing you to inspect, modify, and contribute to the core logic. Check out our latest implementations:

• GEMM + Amax: Optimized FP8 matrix multiplication with absolute maximum calculation.
• GEMM + SwiGLU: High-performance implementation of the SwiGLU activation fused with GEMM.
• Grouped GEMM + GLU: Unified grouped GEMM GLU API supporting dense and discrete MoE weight layouts.
• Grouped GEMM + dGLU: Unified grouped GEMM dGLU backward API supporting dense and discrete MoE weight layouts.
• Grouped GEMM + SwiGLU: Legacy contiguous-only grouped GEMM SwiGLU API.
• Grouped GEMM + dSwiGLU: Legacy contiguous-only grouped GEMM dSwiGLU API.
• Discrete Grouped GEMM + SwiGLU: Per-expert-pointer SwiGLU grouped GEMM for MoE workloads without weight packing.
• Discrete Grouped GEMM + dSwiGLU: Per-expert-pointer dSwiGLU backward grouped GEMM for MoE workloads without weight packing.
• Grouped GEMM + Quant: Legacy dense-only grouped GEMM quant API for MoE FC2/dFC1 workloads.
• Grouped GEMM + Quant (Unified): Unified grouped GEMM quant API with per-row gating for MoE FC2/dFC1 workloads.
• NSA: Native Sparse attention as described in the Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention.

Key Features

• Unified Graph API: Create reusable, persistent cudnn_frontend::graph::Graph objects to describe complex subgraphs.
• Ease of Use: Simplified C++ and Python bindings (via pybind11) that abstract away the boilerplate of the backend API.
• Performance: Built-in autotuning and support for the latest NVIDIA GPU architectures.

Installation

🐍 Python

The easiest way to get started is via pip:

pip install nvidia_cudnn_frontend

Requirements:

• Python 3.8+
• NVIDIA driver and CUDA Toolkit

⚙️ C++ (Header Only)

Since the C++ API is header-only, integration is seamless. Simply include the header in your compilation unit:

#include <cudnn_frontend.h>

Ensure your include path points to the include/ directory of this repository.

Building from Source

If you want to build the Python bindings from source or run the C++ samples:

1. Dependencies

• python-dev (e.g., apt-get install python-dev)
• Dependencies listed in requirements.txt (pip install -r requirements.txt)

2. Python Source Build

pip install -v git+https://github.com/NVIDIA/cudnn-frontend.git

Environment variables CUDAToolkit_ROOT and CUDNN_PATH can be used to override default paths.

3. C++ Samples Build

mkdir build && cd build
cmake -DCUDNN_PATH=/path/to/cudnn -DCUDAToolkit_ROOT=/path/to/cuda ../
cmake --build . -j16
./bin/samples

Documentation & Examples

• Developer Guide: Official NVIDIA Documentation
• C++ Samples: See samples/cpp for comprehensive usage examples.
• Python Samples: See samples/python for pythonic implementations.

🤝 Contributing

We strictly welcome contributions! Whether you are fixing a bug, improving documentation, or optimizing one of our new OSS kernels, your help makes cuDNN better for everyone.

1. Check the Contribution Guide for details.
2. Fork the repo and create your branch.
3. Submit a Pull Request.

Debugging

To view the execution flow and debug issues, you can enable logging via environment variables:

# Log to stdout
export CUDNN_FRONTEND_LOG_INFO=1
export CUDNN_FRONTEND_LOG_FILE=stdout

# Log to a file
export CUDNN_FRONTEND_LOG_INFO=1
export CUDNN_FRONTEND_LOG_FILE=execution_log.txt

Logging Levels:

• CUDNN_FRONTEND_LOG_INFO=0: No logging
• CUDNN_FRONTEND_LOG_INFO=1: Full logging with tensor dumps
• CUDNN_FRONTEND_LOG_INFO=10: Basic logging (safe for CUDA graph capture)

Alternatively, you can control logging programmatically via cudnn_frontend::isLoggingEnabled().

License

This project is licensed under the MIT License.