References: http://blog.csdn.net/u013294888/article/details/56666023 steps:
No. 0 Step: Close secure boot; This step is the most critical, otherwise you will not be able to install ....
First step: Install Nvidia graphics driver
First, disable open source drivers that may cause problems, edit/etc/modprobe.d/blacklist.conf;
sudo gedit/etc/modprobe.d/blacklist.conf
Add Content:
Blacklist AMD76X_EDAC
Blacklist VGA16FB
Blacklist nouveau
Blacklist NVIDIAFB
Blacklist RIVATV//may not need so much
Uninstall clean all installed nvidia drivers:
sudo apt-get remove–purge nvidia-*
Add the driver source by executing the following command:
sudo add-apt-repository Ppa:graphics-drivers/ppa
sudo apt-get update
Below enter tty1 mode execution:
Recommended CTRL+ALT+F1 switch to tty1 execution:
Close LIGHTDM
sudo service LIGHTDM stop
or input: sudo/etc/
sudo apt-get install nvidia-375 nvidia-settings nvidia-prime
sudo nvidia-xconfig//cannot be performed
sudo apt-get install Mesa-common-dev//installation of missing libraries
sudo apt-get install Freeglut3-dev
sudo update-initramfs-u
sudo reboot
Restart should not encounter the problem of circular login;
This is optional, I have no problem:
After the reboot, the entry is still in the tty1 mode:
And then:
SUDO/ETC/INIT.D/LIGHTDM restart
Back to the desktop system.
Second installation CUDA8.0:
I used the. Run installation method,
cd/download/
sudo chmod +x cuda_8.0.61_375.26_linux.run
sudo./cuda_8.0.61_375.26_linux.run
Then install according to the command line prompt:
Click the Q exit clause to browse or press the blanks until the end of the clause, enter accept acceptance clause;
Input n does not install Nvidia image driver, the front has been installed good driver;
Enter Y to install CUDA 8.0 Toolkit;
Carriage return Confirmation Cuda the default installation path (/usr/local/cuda-8.0);
Enter Y to install Cuda 8.0 Samples;
Enter CUDA 8.0 Samples installation path:/home/username/cuda/samples (choose your favorite directory for installation).
At this point, the CUDA8.0 has also been installed.
Then: sudo reboot
After the reboot, enter the CUDNN installation.
Third installation CUDNN:
cd/download/
Tar zxvf cudnn-8.0-linux-x64-v5.1.tgz
CD cuda/
sudo cp include/cudnn.h/usr/local/include/
sudo cp lib64/libcudnn.*/usr/local/lib
Here's how to set up a soft connection:
sudo ln-sf/usr/local/lib/libcudnn.so.5.1.10/usr/local/lib/libcudnn.so.5
sudo ln-sf/usr/local/lib/libcudnn.so.5/usr/local/lib/libcudnn.so
sudo ldconfig-v
To set environment variables:
Add CUDA environment variables to/etc/profile:
sudo gedit/etc/profile
Add the following 2 sentences at the end:
Export ld_library_path=/usr/local/cuda-8.0/lib64/:/usr/local/cuda/lib64:ldlibrarypathexportpath=/usr/local/ Cuda−8.0/bin:/usr/local/cuda/bin:path
After saving, execute the following command to have the environment variable take effect immediately
Source/etc/profile
Also need to add Lib Library path: Add File cuda.conf in/etc/ld.so.conf.d/, as follows
sudo ldconfig
Next, look at your video card situation:
Lspci | Grep-i VGA
Lspci | Grep-i nvidia
Nvcc-v
Where: Disable Nouveau kernel driver, the official website to the method is to add a nvidia-install-xxx.conf file under/etc/modprobe.d/, remove it when disabling.
Then, there is a relatively simple way to switch the graphics drive is in the System-setting software & updates, the last column, additional Drivers, where to switch graphics driver, and then restart the effective test.
Then remember to install cuda8.0, official website: https://developer.nvidia.com/cuda-downloads
To select Runfile, do not use the Deb file .... Say much is the tear ....
The test method is to make the samples and then go to release./devicequery Display video card information