EXTRA:
Some cool commands: nvidia-smi, neofetch
*** THIS IS THE LAST STEP AFTER INSTALLING NVIDIA Drivers, CUDA, cuDNN ***
cd /tmp
curl -O https://repo.anaconda.com/archive/Anaconda3-5.2.0-Linux-x86_64.sh
bash Anaconda3-5.2.0-Linux-x86_64.sh
-Yes add to bashrc
source ~/.bashrc
conda list
conda create --name tfgpu
conda env list
source activate tfgpu
conda install -c anaconda tensorflow-gpu
conda remove -n tfgpu --all
Installing Tensorflow GPU
sudo apt install neofetch
neofetch
***************************************************************************
https://github.com/markjay4k/Install-Tensorflow-on-Ubuntu-17.10-/blob/master/Tensorflow%20Install%20instructions.ipynb
https://www.youtube.com/watch?v=vxjbL5iN1XY
Overview
Step 1: Update your GPU driver (should be higher than version 390)
Step 2: Install the CUDA Toolkit version 9.0 (with all the patches)
Step 3: Install CUDNN 7.0.5
Step 4: Install Tensorflow GPU
Step 5: Test it!
My Case:
Step 1: NVIDIA® GPU drivers —CUDA 9.0 requires 384.x or higher.
GPU Driver:
RTX 2070
Linux x64 (AMD64/EM64T) Display Driver
Version: 418.43
Release Date: 2019.2.22
Operating System: Linux 64-bit
Language: English (US)
File Size: 101.71 MB
bash /home/rb/Downloads/NVIDIA-Linux-x86_64-418.43.run [error]
sudo bash /home/rb/Downloads/NVIDIA-Linux-x86_64-418.43.run [error]
sudo apt install gcc
sudo bash /home/rb/Downloads/NVIDIA-Linux-x86_64-418.43.run [error]
sudo apt install make
sudo bash /home/rb/Downloads/NVIDIA-Linux-x86_64-418.43.run [error]
***Some 32 bit error**
So
sudo dpkg --add-architecture i386
sudo apt update
sudo apt install build-essential libc6:i386
disable the default nouveau Nvidia driver
sudo bash -c "echo blacklist nouveau > /etc/modprobe.d/blacklist-nvidia-nouveau.conf"
sudo bash -c "echo options nouveau modeset=0 >> /etc/modprobe.d/blacklist-nvidia-nouveau.conf"
this will create a file named blacklist-nvidia-nouveau.conf in /etc/modprobe.d with contents:
blacklist nouveau
options nouveau modeset=0
Can confirm using:
cat /etc/modprobe.d/blacklist-nvidia-nouveau.conf
Then,
sudo update-initramfs -u
The update-initramfs script manages your initramfs images on your local box.
It keeps track of the existing initramfs archives in /boot.
There are three modes of operation create, update or delete. ...
At boot time, the kernel unpacks that archive into RAM disk, mounts and uses it as initial root file system.
Then
sudo reboot
In order to install new Nvidia driver we need to stop the current display server.
The easiest way to do this is to change into runlevel 3 using the telinit command.
After executing the following linux command the display server will stop,
therefore make sure you save all your current work ( if any ) before you proceed:
sudo telinit 3
Hit CTRL+ALT+F1 and login with your username and password to open a new TTY1 session.
Then:
sudo bash /home/rb/Downloads/NVIDIA-Linux-x86_64-418.43.run
The distribution-provided pre-install script failed!
Are you sure you want to continue? -> CONTINUE INSTALLATION
Would you like to run the nvidia-xconfig utility? -> YES
sudo reboot
After reboot you should be able to start NVIDIA X Server Settings app from the Activities menu.
https://linuxconfig.org/how-to-install-the-nvidia-drivers-on-ubuntu-18-04-bionic-beaver-linux
https://www.tensorflow.org/install/gpu
Command:
nvidia-smi
Step 2:
CUDA Toolkit:
https://developer.nvidia.com/cuda-downloads?target_os=Linux&target_arch=x86_64&target_distro=Ubuntu&target_version=1804&target_type=deblocal
sudo apt-key add /var/cuda-repo-10-1-local-10.1.105-418.39/7fa2af80.pub
sudo apt-get update
sudo apt-get install cuda
Error:
The following packages have unmet dependencies:
cuda : Depends: cuda-10-1 (>= 10.1.105) but it is not going to be installed
Means you are missing cuda-10-1 package
sudo apt-get install cuda-10-1
but on and on packages keep missing
So better,
Solution:
sudo apt-get install aptitude
sudo aptitude install cuda
Maybe aptitude crawls back, gets all the missing dependencies and installs it for you and then finally installs the main package.
Update your PATH variable
sudo gedit ~/.bashrc
export PATH=/usr/local/cuda-10.1/bin${PATH:+:$PATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-10.1/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
[Yes 2 }} at the end]
will basically add /usr/local/cuda-10.1/bin to path
source ~/.bashrc
or
. ~/.bashrc
Step 3: cuDNN install
Download cuDNN Library for Linux
https://developer.nvidia.com/cudnn
tar -xzvf cudnn-10.1-linux-x64-v7.5.0.56.tgz
sudo cp cuda/include/cudnn.h /usr/local/cuda/include
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*
Done cuDNN.
source ~/.bashrc
Restart terminal
source activate tfgpu
conda install jupyter notebook
jupyter notebook
import tensorflow as tf
print(tf.__version__)
hello = tf.constant('hello tensorflow')
with tf.Session() as sess:
print(sess.run(hello))
Output:
1.12.0
b'hello tensorflow'
