/server-usage

The usage of Vislab server

Server-usage

Vislab is committed to applying deep learning models in computer vision area.

Vislab has three GPU servers under Worcester Polytechnic Insititute sub net, which can provide strong computational ability for deep learning training and test.

Log in

There are two options for connecting to Vislab server

  • Connect to WPI campus Wifi(on campus)
  • Connect to WPI VPN(remote), you can download here and find the configuration here
ssh <usrname>@<vislab-x>.wpi.edu
password

Usage

View all existing docker image

docker image ls

Download new docker image

View all docker image in docker hub.

Follow the instruction to pull the image in the server.

docker pull <repository>:<tag>

Activate a docker container

docker run --gpus=all -it --name <container name> --shm-size 8G --mount type=bind,source=<absolute path>,target=/workspace <repository>:<tag>

User name must be included in container name.

Exit a container

exit

Docker will stop if you run the exit command.

Basic syntax for container

Check all containers

docker container ls -a

Start a container

docker start <container name>

Attach a container

docker attach <container name>

Stop a container

docker stop <container name>

Remove a container

docker rm <container name>

GPU usage

Check available GPUs

nvidia-smi

Run your code

CUDA_VISIBLE_DEVICES=<GPU-index1>, <GPU-index2> <script>

GPU index starts from 0