docker-compose-gpu.yml - do the libraries used for GPU support Blackwell / SM-120 / RTX 5090 / PyTorch 2.7.0+ / CUDA 12.8+ ?

[jhsmith409]

docker-compose-gpu.yml - do the libraries used for GPU support Blackwell / SM-120 / RTX 5090 / PyTorch 2.7.0+ / CUDA 12.8+ ?

[dosubot[bot]]

Hi @jhsmith409! I'm Dosu and I’m helping the ragflow team.

The docker-compose-gpu.yml file only provides basic NVIDIA GPU passthrough and does not explicitly install or configure libraries for Blackwell, SM-120, RTX 5090, PyTorch 2.7.0+, or CUDA 12.8+. The file is not actively maintained, and supporting the latest hardware or CUDA/PyTorch versions would require significant updates to the Dockerfile and package management. Current GPU support is superficial and does not guarantee compatibility with modern NVIDIA hardware or software stacks reference.

Also, maintainers recommend using an embedding server with GPUs rather than using GPUs for document parsing in RAGFlow, as parsing is still CPU-bound and third-party inference frameworks are considered more stable reference.

If you run into errors, make sure to define all required environment variables in your .env file or directly in the compose file reference.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}   

[ZhenhangTung]

@jhsmith409 Hi, will double check with our eng team about your provided config.

[ZhenhangTung]

@jhsmith409 Hi, in summary, yes we do. If you encounter any problem, feel free to reply here.