Ubuntu 16.04 CUDA 8.0

1. Windows 10 & Ubuntu 16.04 Dual-Boot
(Ubuntu 16.04.1 LTS)[http://www.ubuntu.com/download/desktop]
Universal USB Installer 
Boot from USB and install ubuntu
  - Mount point / 200GB
  - Mount point /home 2TB
Reboot after successfully installed

Reference: http://www.tecmint.com/install-ubuntu-16-04-alongside-with-windows-10-or-8-in-dual-boot/

2.Actions for installing Nvidia Driver
Update the System:

  sudo apt-get update sudo apt-get upgrade

Blacklist Nouveau:

  vim /etc/modprobe.d/blacklist
  
  blacklist amd76x_edac #this might not be required for x86 32 bit users.
  blacklist vga16fb
  blacklist nouveau
  blacklist rivafb
  blacklist nvidiafb
  blacklist rivatv
  options nouveau modeset=0
  
Remove all the nvidia* packages: sudo apt-get remove --purge nvidia*
Regenerate kernel: sudo update-initramfs -u
Reboot and press Ctrl + Alt + F1 before login to go shell
Change to run level 3: sudo init 3
Disable lightdm: sudo service lightdm stop
Install Nvidia driver:

  sudo add-apt-repository ppa:graphics-drivers/ppa
  sudo apt-get update
  sudo apt-get install nvidia-367

Configure Xorg: sudo nvidia-xconfig
Reboot

References:
https://devtalk.nvidia.com/default/topic/940287/gtx-1080-newest-drivers-cuda-toolkit-8-0-27-rc-installed-nothing-working-/
http://yangcha.github.io/GTX-1080/

3.Actions for installing CUDA8.0
Download Installers for Linux Ubuntu 16.04 x86_64 (https://developer.nvidia.com/cuda-release-candidate-download)
  * runfile (local)
  - Base Installer
  - Patch 1 (Released Aug 8, 2016)

Follow the guide here(http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#runfile) for runfile installation

  sudo ./cuda_8.0.27_linux.run --override

  -------------------------------------------------------------
  Do you accept the previously read EULA? (accept/decline/quit): accept
  You are attempting to install on an unsupported configuration. Do you wish to continue? ((y)es/(n)o) [ default is no ]: y
  Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 352.39? ((y)es/(n)o/(q)uit): n
  Install the CUDA 7.5 Toolkit? ((y)es/(n)o/(q)uit): y
  Enter Toolkit Location [ default is /usr/local/cuda-7.5 ]:
  Do you want to install a symbolic link at /usr/local/cuda? ((y)es/(n)o/(q)uit): y
  Install the CUDA 7.5 Samples? ((y)es/(n)o/(q)uit): y
  Enter CUDA Samples Location [ default is /home/kinghorn ]: /usr/local/cuda-7.5
  Installing the CUDA Toolkit in /usr/local/cuda-7.5 ...
  Finished copying samples.

  * use --override
  * no for "Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 XXX.XX?"

Install patch: 

  sudo cuda_8.0.27.1_linux.run

Follow the guide here(http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#post-installation-actions) after installed CUDA

Refresh the shared libs cache
sudo ldconfig /usr/local/cuda-8.0/lib64

4. Install cuDNN 8.0
Download cuDNN here(https://developer.nvidia.com/rdp/cudnn-download)
cd cuda
sudo cp -P lib64/* /usr/local/cuda/lib64/
sudo cp -P include/cudnn.h /usr/local/cuda/include/

5. Theano Config
[global]
floatX = float32
device = gpu

[lib]
cnmem = 0.9

Test theano by using commands

  import theano.sandbox.cuda
  theano.sandbox.cuda.use("gpu0")
  
You should see a message like this:

  Using gpu device 0: GeForce GTX 1080 (CNMeM is enabled with initial size: 90.0% of memory, cuDNN 5105)

Run check_blas.py to see test results:

  python `python -c "import os, theano; print(os.path.dirname(theano.__file__))"`/misc/check_blas.py


Troubleshooting:
1. "Input Signal Out of Range. Change Settings to 1920 x 1080"
  Probably need to change monitor 
2. Failed to compile cuda_ndarray.cu: libcublas.so.8.0: cannot open shared object file
   or 'could not create cuDNN handle: CUDNN_STATUS_INTERNAL_ERROR'
  sudo ldconfig /usr/local/cuda-8.0/lib64