openeuler/cuda repository overview

⁠Quick reference

• The official CUDA docker image.

• Maintained by: openEuler CloudNative SIG⁠.

• Where to get help: openEuler CloudNative SIG⁠, openEuler⁠.

⁠CUDA | openEuler

Current CUDA docker images are built on the openEuler⁠. This repository is free to use and exempted from per-user rate limits.

A heterogeneous computing platform for general-purpose parallel computing that provides layered APIs and high-level libraries to help you quickly build high-performance computing applications and AI services based on NVIDIA GPUs.

Learn more about on CUDA Document⁠.

⁠Supported tags and respective Dockerfile links

The tag of each cuda docker image is consist of the complete software stack version. The details are as follows

⁠Usage

In this usage, users can select the corresponding {Tag} and container startup options based on their requirements.

• Pull the openeuler/cudas image from docker

  docker pull openeuler/cuda:{Tag}
  

  Copy

• Start a cann instance

  docker run \
      --name my-cuda \
      --gpus all \
      -it openeuler/cuda:{Tag} bash
  

  Copy

• Container startup options

  Option	Description
  --name my-cuda	Names the container my-cuda.
  --gpus all	The specified container can access all GPU devices, you can also specify a specific GPU, such as --gpus '"device=0,1"'
  -it	Starts the container in interactive mode with a terminal (bash).
  openeuler/cuda:{Tag}	Specifies the Docker image to run, replace {Tag} with the specific version or tag of the openeuler/cuda image you want to use.

• View container running logs

  docker logs -f my-cuda
  

  Copy

• To get an interactive shell

  docker exec -it my-cuda /bin/bash
  

  Copy

⁠Question and answering

If you have any questions or want to use some special features, please submit an issue or a pull request on openeuler-docker-images⁠.