Initialize VM Instance on Google Cloud Platform
A guide to install CUDA, cuDNN, Anaconda, OpenCV, Pytorch.
-
My GPU: NVIDIA Tesla T4
-
My System: Ubuntu 18.04 LTS
Create VM Instance
Follow this instruction.
Preparing
Get some fundamental packages and updates.
sudo apt-get update
sudo apt-get upgrade
And:
sudo apt-get install -y make build-essential libssl-dev zlib1g-dev libbz2-dev libreadline-dev libsqlite3-dev wget curl llvm libncurses5-dev libncursesw5-dev xz-utils tk-dev valgrind cmake unrar gfortran python3-pip python3-dev python3-wheel swig git git-core htop g++ freeglut3-dev libx11-dev libxmu-dev libxi-dev libglu1-mesa libglu1-mesa-dev
Original python version is python3.5, change to python3.7:
wget https://www.python.org/ftp/python/3.7.3/Python-3.7.3.tgz
tar zxvf Python-3.7.3.tgz
cd Python-3.7.3
./configure --with-ssl
make
sudo make install
cd ..
Check if successfully instaled:
python3
Get packages.
sudo apt-get install python3-pip python3-setuptools
sudo pip3 install --upgrade pip
sudo pip3 install numpy scipy matplotlib pandas seaborn sklearn lightgbm xgboost tqdm
Install CUDA
Find your package at Nvidia offical website.
sudo wget http://us.download.nvidia.com/tesla/418.87/NVIDIA-Linux-x86_64-418.87.00.run
sudo chmod +x NVIDIA-Linux-x86_64-418.87.00.run
sudo ./NVIDIA-Linux-x86_64-418.87.00.run
And:
sudo apt-get update
sudo apt-get upgrade
Check if successfully installed.
nvidia-smi
My result:
Driver Version: 418.87.00
CUDA Version: 10.1
Install cuDNN
Create an account and download file.
sudo dpkg -i libcudnn7_7.6.3.30-1+cuda10.1_amd64.deb
Install Anaconda
Find your package at Anaconda offical website.
wget https://repo.anaconda.com/archive/Anaconda3-2019.07-Linux-x86_64.sh
bash Anaconda3-2019.07-Linux-x86_64.sh
And:
source ~/.bashrc
Set passward:
vim generateSha.py
Then encrypt:
from notebook.auth import passwd
print(passwd('Your Passwd'))
Run and save the results
python3 generateSha.py
Then follow the instructions.
jupyter notebook --generate-config
vim /home/usr/.jupyter/jupyter_notebook_config.py
Add these to file:
c = get_config()
c.NotebookApp.ip = '*'
c.NotebookApp.open_browser = False
c.NotebookApp.port = <Port Number>
c.NotebookApp.password = u'sha1:your result'
Exit file and open notebook with passwd.
Can't open website?
-
Use http request, not https.
-
Close all vpn.
-
Close all chrome plugin which may block the website.
Install OpenCV
sudo pip3 install opencv-python
pip install opencv-python
Install Pytorch
Use conda install or other means.
conda update --all
And:
conda install pytorch torchvision cudatoolkit=10.0 -c pytorch
Check if successfully installed.
python3
import torch
torch.__version__
My result: '1.2.0'
Warning
Remember to shut down VM instance or you will be charged even if you are not using it.