ubuntu14.04+cuda8.0+cudnn6.0+python3+tensorflow1.0源码安装

tensorflow源码安装过程

   1. 安装cuda8.0和cudnn6.0（http://blog.csdn.net/disen10/article/details/79279969） 

   2. 安装python3

       sudo apt-get install python3

       安装pip3:sudo apt-get install python3-pip python3-setuptools python3-numpy cython3

       修改用户默认python版本:打开~/.bashrc，添加alias python='/usr/bin/python3.4'

   3. 安装bazel(https://docs.bazel.build/versions/master/install-ubuntu.html#install-on-ubuntu) 

   4. 安装第三放依赖库

 sudo apt-get install python-numpy swig python-dev python-wheel python-setuptools

 sudo apt-get install git

   5. 安装tensorflow

 源码下载：git clone --recursive https://github.com/tensorflow/tensorflow.git

  cd tensorflow

 git checkout r1.0（检查版本号)

  sudo apt-get install libcupti-dev（配置环境)

  sudo apt-get upgrade(升级驱动)   

 cd tensorflow

 ./configure(配置tensorflow)

 bazel build --config=opt --config=cuda //tensorflow/tools/pip_package:build_pip_package(build the pip package)

 bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg(create .whl file)

  sudo pip3 install /tmp/tensorflow_pkg/(your file name).whl（install tensorflow)

   6. 测试tensorflow:

 cd ~

 在terminal输入python3：

   >>> import tensorflow as tf  

  >>> hello = tf.constant('Hello, TensorFlow!')  

  >>> sess = tf.Session()  

   >>> print(sess.run(hello))