Skip to content

Software Deployment

Software deployment on DGX-2 is based on containers. NVIDIA provides a wide range of prepared Docker containers with a variety of different software. A user can easily download these containers and use them directly on the DGX-2.

The catalog of all container images can be found on NVIDIA site. Supported software includes:

  • TensorFlow
  • MATLAB
  • GROMACS
  • Theano
  • Caffe2
  • LAMMPS
  • ParaView
  • ...

Running Containers on DGX-2

NVIDIA expects usage of Docker as a containerization tool, but Docker is not a suitable solution in a multiuser environment. For this reason, the Singularity container solution is used.

Singularity can be used very similar to Docker, the only change is a rewrite of an image URL address. For example, original command for Docker docker run -it nvcr.io/nvidia/theano:18.08 should be rewritten to singularity shell docker://nvcr.io/nvidia/theano:18.08. More about Singularity here.

Info

The --nv Singularity switch is used by default on DGX-2.

For fast container deployment, all images are cached after first use in lscratch directory. This behavior can be changed by SINGULARITY_CACHEDIR environment variable, but the start time of container will increase significantly.

Comments