Detailed explanation of Ubuntu16.04 installation of nvidia driver + CUDA+cuDNN tutorial

Preparatory work

1. Check whether GPU supports CUDA

lspci | grep -i nvidia

2. View the Linux version

uname -m && cat /etc/*release

nvidia driver

1. Uninstall the original N card driver first

#for case1: original driver installed by apt-get:
sudo apt-get remove --purge nvidia*
#for case2: original driver installed by runfile:
sudo chmod +x *.run
sudo ./NVIDIA-Linux-x86_64-384.59.run --uninstall

2. Disable the nouveau driver

sudo gedit /etc/modprobe.d/blacklist.conf

At the end of the text, add:

blacklist nouveau
options nouveau modeset=0

Then execute:

sudo update-initramfs -u

After restarting, you can see if nouveau is running:

lsmod | grep nouveau  # 没输出代表禁用生效

3. Disable the X-Window service

sudo service lightdm stop # This closes the graphical interface

Press Ctrl-Alt+F1 to enter the command line interface and enter the user name and password to log in.

4. Install the driver from the command line

Driver website https://www.nvidia.cn/Download/index.aspx? lang = cn

# Give drive run File gives execute permission: 
sudo chmod +x NVIDIA-Linux-x86_64-384.59.run
# The following parameters are very important and cannot be omitted: 
sudo ./NVIDIA-Linux-x86_64-384.59.run  In fact, in fact, the no-opengl-files

5. Driver test:

nvidia-smi # If the information list of GPU is listed, the driver installation is successful
nvidia-settings # If the setting dialog box pops up, it also means that the driver is successfully installed

CUDA Installation

1. Download the CUDA for the corresponding system

Website http://developer.nvidia.com/cuda-downloads Select runfile Installation

Step 2 Install

sudo sh cuda_<version>_linux.run

After starting the installation, you need to read the instructions. You can read them directly using Ctrl + C, or you can read them slowly using the space bar. The following are the installation options:

 (If you agree to the terms, you must agree to continue the installation) 
accept/decline/quit: accept
 (Do not install the driver here, because the latest driver has been installed, otherwise the old version of the graphics card driver may be installed, resulting in repeated login.) 
Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 410.48?
(y)es/(n)o/(q)uit: n
Install the CUDA 10.0 Toolkit? (Is it installed CUDA 10  , which must be installed here) 
(y)es/(n)o/(q)uit: y
Enter Toolkit Location (Installation path, use default, just enter directly) 
[ default is /usr/local/cuda-10.0 ]: 
Do you want to install a symbolic link at /usr/local/cuda? (Agree to create soft link) 
(y)es/(n)o/(q)uit: y
Install the CUDA 10.0 Samples? (You don't need to install the test, you have it yourself.) 
(y)es/(n)o/(q)uit: n
Installing the CUDA Toolkit in /usr/local/cuda-10.0 ... (Start installation) 

3. Configure environment variables

sudo gedit ~/.bashrc

Add at the end

export PATH=/usr/local/cuda-8.0/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH

Update

source ~/.bashrc

4. Verify the installation

View cuda version

nvcc -V
CUDA Sample Test: 
# Compile and test equipment  deviceQuery : 
cd /usr/local/cuda-8.0/samples/1_Utilities/deviceQuery
make
./deviceQuery
# Compile and test bandwidth  bandwidthTest : 
cd ../bandwidthTest
make
./bandwidthTest

If the final result of both tests is Result = PASS, CUDA is installed successfully.

* 5. CUDA Uninstall

Enter on the command line

sudo apt-get remove cuda 
sudo apt-get autoclean
sudo apt-get remove cuda*

Then switch to usr/local/in the directory

cd /usr/local/
sudo rm -r cuda-9.1

cuDNN Installation

Download the corresponding version cuDNN https://developer.nvidia.com/cudnn

tar xvzf cudnn-9.2-linux-x64-v7.1
sudo cp -P cuda/include/cudnn.h /usr/local/cuda/include
sudo cp -P cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*
sudo ldconfig

Reference

https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#post-installation-actions

Summarize