installGPU.md

GPU Setup:

 

Step 1: Initialization

user@machine:~$ LAMBDA_REPO=$(mktemp) && wget -O${LAMBDA_REPO} https://lambdalabs.com/static/misc/lambda-stack-repo.deb && sudo dpkg -i ${LAMBDA_REPO} && rm -f ${LAMBDA_REPO} && sudo apt-get update && sudo apt-get install -y lambda-stack-cuda

Step 2: Reboot / Restart

user@machine:~$ sudo reboot # Run after 3-4 minutes.

Step 3: Nvidia Initialization

user@machine:~$ sudo docker build -t lambda-stack -f Dockerfile.$(lsb_release -cs) git://github.com/lambdal/lambda-stack-dockerfiles.git

Step 4: Nvidia Installation

user@machine:~$ LAMBDA_REPO=$(mktemp) && wget -O${LAMBDA_REPO} https://lambdalabs.com/static/misc/lambda-stack-repo.deb && sudo dpkg -i ${LAMBDA_REPO} && rm -f ${LAMBDA_REPO} && sudo apt-get update && sudo apt-get --yes upgrade && sudo apt-get install --yes --no-install-recommends lambda-server && sudo apt-get install --yes --no-install-recommends nvidia-440 libcuda1-440 nvidia-opencl-icd-440 && sudo apt-get install --yes --no-install-recommends lambda-stack-cuda

Step 5: Reboot / Restart (Again)

user@machine:~$ sudo reboot # Run after 3-4 minutes.

 

Reference: https://lambdalabs.com/lambda-stack-deep-learning-software