Build Deep Learning Environment on Ubuntu 16.04 (NVIDIA cuDNN and Microsoft CNTK)

Ubuntu_16.04_NVIDA_cuDNN_CNTK.md

Download Proprietary NVIDIA driver for "GeForce GTX 970"

wget NVIDIA-Linux-x86_64-384.90.run

disable nouveau driver

How to disable nouveau kernel driver on Ubuntu 16.04 [ https://askubuntu.com/a/868209/497162 ]

sudo vim /etc/modprobe.d/nvidia-graphics-drivers.conf
sudo vim  /etc/modprobe.d/nvidia-installer-disable-nouveau.conf
sudo apt-get remove nvidia\*
sudo apt-get install dkms build-essential linux-headers-generic
sudo vim /etc/modprobe.d/blacklist.conf
echo options nouveau modeset=0 | sudo tee -a /etc/modprobe.d/nouveau-kms.conf
sudo vim /etc/modprobe.d/nouveau-kms.conf
sudo update-initramfs -u
sudo reboot

install proprietary NVidia driver

sudo service lightdm stop
sudo sh NVIDIA-Linux-x86_64-384.90.run
nvidia-smi

Download and Install NVidia CUDA debian file

wget cuda-repo-ubuntu1604-9-0-local_9.0.176-1_amd64.deb
sudo apt-key add /var/cuda-repo-<your-cuda-version>-local/7fa2af80.pub
sudo apt-get update
sudo apt-get install cuda

Download and install NVidia cuDNN

[ https://medium.com/@ishengfang/%E5%A6%82%E4%BD%95%E5%9C%A8ubuntu-17-04%E6%9E%B6%E8%A8%ADdeep-learning-%E7%92%B0%E5%A2%83-ca7377d98d5d ]

tar xvf cudnn-9.0-linux-x64-v7.tgz
cd cuda
sudo cp include/cudnn.h /usr/local/cuda/include/
sudo cp lib64/libcudnn* /usr/local/cuda/lib64/
sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*
sudo reboot

Anaconda 3 Python 3.5 create virutal environment for CNTK-2.2

[ https://medium.com/@ishengfang/%E5%A6%82%E4%BD%95%E5%AE%89%E8%A3%9Dcntk%E4%BD%9C%E7%82%BAkeras%E7%9A%84backend%E4%B8%A6%E8%88%87tensorflow%E6%AF%94%E8%BC%83-4a21b0f016d6 ]

conda create -n CNTKpy35 python=3.5 anaconda
source activate CNTKpy35
pip install https://cntk.ai/PythonWheel/GPU/cntk-2.2-cp35-cp35m-linux_x86_64.whl
ipython3

In [1]: import cntk as C

In [2]: print(C.__version__)
2.2

In [3]: C.device.try_set_default_device(C.device.gpu(0))
Out[3]: True