Commands required to setup a GCE instance to run Distributed Tensorflow

distirbuted_tensorflow_setup.txt

# Install pip
wget https://bootstrap.pypa.io/get-pip.py	
sudo python get-pip.py

#  Install CUDA 9 
curl -O http://developer.download.nvidia.com/compute/cuda/rep
os/ubuntu1604/x86_64/cuda-repo-ubuntu1604_9.0.176-1_amd64.deb
sudo dpkg -i cuda-repo-ubuntu1604_9.0.176-1_amd64.deb
sudo apt-get update
sudo apt-get install cuda

# Install CUDnn 7 library
# The driver has been uploaded to a GCS bucket for easy access
gsutil cp gs://lib-driver .
sudo dpkg -i libcudnn7_7.0.3.11-1+cuda9.0_amd64.deb

# Install TensorFlow 1.8
pip install tensorflow-gpu

# Run the ResNet50 model
git clone https://github.com/tensorflow/models.git
export PYTHONPATH="$PYTHONPATH:/home/anjalisridhar/models"
sudo  pip install -r models/official/requirements.txt
cd models/official/resnet

# GPU = 8
python imagenet_main.py  --batch_size=1024  --model_dir=gs://singlenode-gcp-benchmarks/demo-v100-numgpus8_r2 --dtype=fp16 --data_dir=/mnt/disks/imagenet/imagenet/combined --num_gpus=8

# Spin up Tensorboard using the Google Cloud shell
tensorboard —log_dir=gs://singlenode-gcp-benchmarks/demo_model_dir_numgpus8