Ubuntu Deep learning Environment Building Tensorflow+pytorch

Current Computer Configuration: Ubuntu 16.04 + GTX1080 Graphics

Configuring a deep learning environment, using Tsinghua Source to install a Miniconda environment is a very good choice. In particular, today found Conda install-c Menpo opencv3 A command can be smoothly installed on the OPENCV, before their own time also encountered a lot of errors. Conda installation of the TensorFlow and pytorch two kinds of framework is also very convenient, for not good at source code compilation I was the best choice right.

So the general process is: Install the graphics driver-install cuda 8.0--install cudnn--install miniconda--install various calculation packages

The command is as follows:

Installation drive

sudo add-apt-repository Ppa:graphics-drivers/ppa

sudo apt-get update

sudo apt-get install nvidia-367

sudo apt-get install Mesa-common-dev

sudo apt-get install Freeglut3-dev

Reboot system for GTX1080 graphics driver to take effect

Download Cuda 8.0 Run file

sudo sh cuda_8.0.61_375.26_linux.run

Q Fast Forward skip, prompts whether to install XXXX select n

Configuring environment variables to ~/.BASHRC

Export path=/usr/local/cuda-8.0/bin/: $PATH

Export ld_library_path= "/usr/local/cuda-8.0/lib64:/usr/local/cuda-8.0/extras/cupti/lib64"

Installation CUDNN is relatively simple, after extracting the corresponding files copied to the corresponding Cuda directory can be

sudo cp cudnn.h/usr/local/cuda/include/#复制头文件

sudo cp lib*/usr/local/cuda/lib64/#复制动态链接库

sudo rm-rf libcudnn.so libcudnn.so.6 #删除原有动态文件

sudo ln-s libcudnn.so.6.0.21 libcudnn.so.6 #生成软链接

sudo ln-s libcudnn.so.6 libcudnn.so #生成软链接

Installing Miniconda

Configure Tsinghua Source

Download Miniconda (python3.6)

Bash miniconda3-latest-linux-x86_64.sh

Install version Tensorflow-gpu

Conda install-y tensorflow-gpu==1.4.1

Pytorch Official website installation is very simple not to write

Ubuntu Deep learning Environment Building Tensorflow+pytorch