NVIDIA/dl-cuda-graph-doc

NVIDIA/dl-cuda-graph-doc

Description: A comprehensive guide to using CUDA Graphs effectively with PyTorch, covering CUDA fundamentals, PyTorch integration, Megatron-LM implementations, and practical troubleshooting.

Language: Shell

License: NOASSERTION

Stars: 10

Forks: 3

Open issues: 1

Created: 2026-01-22T04:53:52Z

Pushed: 2026-01-26T14:02:15Z

Default branch: main

Fork: no

Archived: no

README:

CUDA Graph Best Practice For PyTorch

A comprehensive guide to using CUDA Graphs effectively with PyTorch, covering CUDA fundamentals, PyTorch integration, Megatron-LM implementations, and practical troubleshooting.

📚 Documentation

View the documentation at: https://docs.nvidia.com/dl-cuda-graph/

📖 What's Inside

• CUDA Graph Basics: Fundamentals, constraints, and quantitative benefits
• PyTorch CUDA Graphs: Integration, Transformer Engine & Megatron-LM, best practices, and handling dynamic patterns
• Examples: Real-world implementations from MLPerf Training (Llama 3.1 405B, GPT-3 175B, Stable Diffusion v2, RNN-T)
• Troubleshooting: Capture failures, numerical errors, memory issues, and performance debugging

🚀 Quick Start

# Clone and serve locally (auto-installs dependencies)
git clone https://github.com/NVIDIA/dl-cuda-graph-doc.git
cd dl-cuda-graph-doc
./scripts/sphinx-serve.sh

Visit http://127.0.0.1:8000 to view the documentation.

Manual build:

uv sync --group docs
uv run --group docs sphinx-build docs docs/_build/html

📝 Contributing

We welcome contributions! Please see [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines.

📜 License

This project uses dual licensing:

• Documentation (Markdown files): [CC-BY-4.0](LICENSE-DOCS) - Creative Commons Attribution 4.0 International
• Code (Python, shell scripts, configuration): [MIT](LICENSE)

See [THIRD-PARTY-NOTICES.md](THIRD-PARTY-NOTICES.md) for third-party dependencies and their licenses.

🔗 Resources

• CUDA Programming Guide
• PyTorch CUDA Graphs
• Sphinx with NVIDIA Sphinx Theme