Running Containers

The following is an example of running the CUDA container from the NGC registry.

sudo docker run --security-opt label=type:nvidia_container_t --rm nvcr.io/nvidia/cuda:10.0-runtime nvidia-smi
Note: To accommodate SELinux, the DGX software stack includes a package (nvidia-container-selinux) that defines a policy for allowing containers to access NVIDIA GPUs. The --security-opt option in the command sets the corresponding label type permitting the specified container to access NVIDIA GPUs. If SELinux is removed or disabled, then the --security-opt option is not needed.