Complete tutorial to set up Ubuntu Desktop on GCP with Tensorflow GPU and Jupyter Notebook
Feel free to get in touch with me on LinkedIn or ask any questions:
/ helmy47
----------------------------------------------------------
In this video, we will set up anaconda-navigator - Jupyter notebook with TensorFlow-GPU with an NVIDIA Tesla V100 on GCP.
The versions we set up are listed below, but you can follow this video to set up any combinations in any cloud platform AWS or Azure:
---
Cuda: 10.0
Cudnn: 7.4
Tensorflow_GPU-1.14.0
Python 3.7
----
Relevant Links:
GCP:
https://console.cloud.google.com/
Google SDK (Make sure to set it up correctly):
https://cloud.google.com/sdk/docs/dow...
VNC Viewer:
https://www.realvnc.com/en/connect/do...
Tensorflow GPU Compatibility table with CUDA, CUDNN & Python:
https://www.tensorflow.org/install/so...
CUDA Website (Access on Linux GCP):
https://developer.nvidia.com/cuda-too...
CUDNN Website (Access on Linux GCP - Requires Login):
https://developer.nvidia.com/rdp/form...
Google SDK Command:
gcloud compute ssh [INSTANCE-NAME-HERE] --project [PROJECT-NAME-HERE] --zone [REGION-NAME-HERE] --ssh-flag "-L 5901:localhost:5901
Creating a Desktop icon for Anaconda:
sudo mkdir /home/USERNAME/bin
sudo nano /home/USERNAME/bin/anaconda.sh
(Follow the video to ensure you don't get all the errors because of the spaces)
export CONDA_PYTHON_EXE=/home/maged_helmy/anaconda3/bin/python export PATH=/home/maged_helmy/anaconda3/bin:$PATH python /home/maged_helmy/anaconda3/bin/anaconda-navigator
/home/maged_helmy/Desktop/anaconda.desktop
[Desktop Entry]
Version=1.0
Type=Application
Name=Anaconda-Navigator
GenericName=Anaconda
Exec=/bin/bash "/home/maged_helmy/bin/anaconda.sh" Icon=/home/maged_helmy/anaconda3/lib/python3.6/site-packages/anaconda_navigator/static/images/anaconda-icon-256x256.png Terminal=false
StartupNotify=true
sudo desktop-file-install anaconda.desktop
sudo chmod +x *.desktop
If anaconda starts in base after "sudo reboot", use the following command:
conda config --set auto_activate_base false
To create an environment with a python version not listed in anaconda-navigator but installed in the system:
conda create --name test_36 python=3.6
Watch video Complete tutorial to set up Ubuntu Desktop on GCP with Tensorflow GPU and Jupyter Notebook online without registration, duration hours minute second in high quality. This video was added by user HelmyCodeCamp 25 January 2020, don't forget to share it with your friends and acquaintances, it has been viewed on our site 715 once and liked it 15 people.