jhunahn/torch repository overview

⁠torch

This repository provides an environment for using pytorch⁠. The pytorch is an open source machine learning framework that accelerates the path from research prototyping to production deployment. It is configured to support NVIDIA Graphic Accelerator, and drivers for GPU configuration as known as CUDA⁠ are installed.

⁠Usage

By appending options for GPU setting, it can be run with a simple command as shown below.

docker run -it --privileged \
                --gpus $device --ipc=host \
                -v `pwd`:/src -w /src \
                --rm jhunahn/torch:$TAGNAME

• The --gpus option is the number of GPUs. If you enter all, all the GPUs on the host are enabled.

⁠X11 forwarding from a docker container

• Set DISPLAY environment variable
• mount volume /tmp/.X11-unix
• shared tmpfs

docker run -it --privileged \
                --gpus '$device,"capabilities=compute,utility,graphics"' \
                --ipc=host --net=host \
                -v `pwd`:/src -w /src \
                -e DISPLAY \
                --device=/dev/dri:/dev/dri \
                -v /tmp/.X11-unix:/tmp/.X11-unix:ro \
                -v /dev/shm:/dev/shm --shm-size=2gb \
                --rm jhunahn/torch:$TAGNAME