Instructions for setting up a GPU instance on Google Cloud

1. Install Google Cloud SDK on your laptop:
   https://cloud.google.com/sdk/

2. Install file by executing the following in the terminal:

    ./google-cloud-sdk/install.sh  
   

3. Initialize gcloud in terminal (https://cloud.google.com/compute/docs/gcloud-compute/?hl=de):  

    gcloud init  
   

4. Request limit increase for instances

5. Put in number of GPUs & CPUs, Size:
   https://console.cloud.google.com/compute/quotas?hl=de&_ga=1.69181649.1124953923.1496188820

6. Create GPU instance:

7. 

8. GPU are only available in certain areas:
   

9. • Choose Boot disk -> Ubuntu 16.04 LTS
   • Choose 30 GB HDD
     

10. Select zone, number of GPUs & CPUs and memory
    

11. Allow HTTP & HTTPS traffic
    

12. Instance is set up
    

13. Connect to instance:  

    gcloud compute config-ssh  
    

1. Or:

   gcloud compute ssh instance-2 --zone us-west1-b
    		ssh –i ~/.ssh/google_compute_engine USERNAME@IP_ADDRESS  
   

2. Install GPU drivers:

   curl –O http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/cuda-repo-ubuntu1604_8.0.61-1_amd64.deb  
   sudo dpkg -i cuda-repo-ubuntu1604_8.0.61-1_amd64.deb  
   sudo apt-get update  
   sudo apt-get install cuda -y  
   

3. Insert lines in bashrc:  

   echo 'export CUDA_HOME=/usr/local/cuda' >> ~/.bashrc  
   echo 'export PATH=$PATH:$CUDA_HOME/bin' >> ~/.bashrc  
   echo 'export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CUDA_HOME/lib64' >> ~/.bashrc  
   source ~/.bashrc  
   

4. Download cuDNN from Nvidia Developer site -> Sign Up required
   • https://developer.nvidia.com/rdp/cudnn-download
   • Save on laptop: https://developer.nvidia.com/compute/machine-learning/cudnn/secure/v5.1/prod_20161129/8.0/cudnn-8.0-linux-x64-v5.1-tgz

5. Open terminal on laptop and copy file to instance:  

   gcloud compute scp cudnn-8.0-linux-x64-v5.1.tgz *INSTANCE_NAME*:~
   

6. Go to terminal on instance and extract & install:

   tar -xzvf cudnn-8.0-linux-x64-v5.1.tgz  
   sudo cp cuda/lib64/* /usr/local/cuda/lib64/  
   sudo cp cuda/include/cudnn.h /usr/local/cuda/include/  
   rm -rf ~/cuda  
   rm cudnn-8.0-linux-x64-v5.1.tgz  
   sudo apt-get update   
   

7. Install python packages:  

   sudo apt-get install python-pip  
   sudo apt-get install python-numpy python-scipy python-matplotlib ipython ipython-notebook  
   sudo apt-get install python-pandas python-sympy 
   

8. Install Anaconda:  

   wget https://repo.continuum.io/archive/Anaconda3-4.4.0-Linux-x86_64.sh  
   bash Anaconda3-4.4.0-Linux-x86_64.sh   
   

9. Set Anaconda as preferred environment:
   • which python should be usr/bin/python

   • Important:

     source .bashrc
     - Generate config file:

     jupyter notebook –generate-config mkdir certs sudo openssl req -x509 -nodes -days 365 -newkey rsa:1024 -keyout mycert.pem -out mycert.pem

• Configure jupyter:

    cd ~/.jupyter/
    vi jupyter_notebook_config.py  
  

• Change username if necessary:

    c = get_config()
    c.IPKernelApp.pylab = 'inline'  # if you want plotting support always in your notebook
    c.NotebookApp.certfile = u'/home/ubuntu/certs/mycert.pem' #location of your certificate file
    c.NotebookApp.ip = '*'
    c.NotebookApp.open_browser = False  #so that the ipython notebook does not opens up a browser by default
    #c.NotebookApp.password = u'sha1:98ff0e580111:12798c72623a6eecd54b51c006b1050f0ac1a62d'  #the encrypted password we generated above
    c.NotebookApp.port = 8888  
  	
    source .bashrc
  

1. Create SSH tunnel:

   ssh -i ~/.ssh/google_compute_engine -L 8899:localhost:8888 USERNAME@35.197.12.149
   jupyter notebook  
   

1. Open Browser and type in token:

   https://localhost:8899/
   

2. Install Keras and Tensorflow:

   pip/conda install keras
   pip/conda install tensorflow-gpu
   conda install opencv
   

3. If GPU process is not running, try to run a keras/tf model first:

   nvidia-smi  
   

Sources / Appendix

https://medium.com/google-cloud/running-jupyter-notebooks-on-gpu-on-google-cloud-d44f57d22dbd
https://cloud.google.com/compute/docs/gpus/add-gpus?hl=de#install-driver-manual
https://cloud.google.com/compute/docs/gcloud-compute/?hl=de

• One line code to create GPU instance:

    gcloud beta compute instances create gpu-inst1 --machine-type n1-standard-2 --zone us-west1-b --accelerator type=nvidia-tesla-k80, count=1 --image-family ubuntu-1604-lts --image-project ubuntu-os-cloud --boot-disk-size 30GB --maintenance-policy TERMINATE --restart-on-failure