Tensorflow1.0+ubuntu14.04+GPU安装

Tensorflow可以通过pip安装也可以通过代码安装（前提是自己已经安装好了显卡驱动http://blog.csdn.net/cugxyy6/article/details/54412014）

1、  pip安装

sudo apt-getinstall python-pip python-dev

pip install tensorflow-gpu
如果安装失败，可能是国内google连不上，可以设置url，通过github来安装
export TF_BINARY_URL=https://ci.tensorflow.org/view/Nightly/job/nightly-matrix-linux-gpu/TF_BUILD_IS_OPT=OPT,TF_BUILD_IS_PIP=PIP,TF_BUILD_PYTHON_VERSION=PYTHON2,label=gpu-linux/lastSuccessfulBuild/artifact/pip_test/whl/tensorflow_gpu-1.0.0-cp27-none-linux_x86_64.whl
 
sudo pip install --upgrade $TF_BINARY_URL
2、  代码安装
1）  首先安装一些依赖库：
sudoapt-get -y update
sudoapt-get -y upgrade
sudoapt-get -y install git python-pip python-dev
sudoapt-get -y install -y libpng12-dev libfreetype6 python-numpy python-scipyipython python-matplotlib build-essential cmake pkg-config libtiff4-devlibjpeg-dev libjasper-dev libgtk2.0-dev libavcodec-dev libavformat-devlibswscale-dev libv4l-dev swig zip python-sklearn python-wheel
2）  安装bazel
A、 安装JDK
$ sudoadd-apt-repository ppa:webupd8team/java
$ sudoapt-get update
$ sudoapt-get install Oracle-java8-installer
B、 添加Bazel URL
$ echo"deb [arch=amd64] http://storage.googleapis.com/bazel-apt stablejdk1.8" | sudo tee /etc/apt/sources.list.d/bazel.list
$ curlhttps://bazel.build/bazel-release.pub.gpg | sudo apt-key add -
C、 安装Bazel
$ sudoapt-get update && sudo apt-get install bazel
$ sudoapt-get upgrade bazel
3）  安装Tensorflow
下载代码
git clonehttps://github.com/tensorflow/tensorflow
安装库
sudo apt-getinstall libcupti-dev
进入tensorflow目录
$ ./configure
按照操作安装，其中Do you wish to build TensorFlow with OpenCL support? [y/N] N
要不然需要安装ComputeCpp编译器(e.g. /usr/local/computecpp)
 
4）  创建pip包并且安装
$ bazel build--config opt //tensorflow/tools/pip_package:build_pip_package
 
    # To build with support for CUDA:
$ bazel build--config opt --config=cuda //tensorflow/tools/pip_package:build_pip_package
 
$bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg
 
# The name ofthe .whl file will depend on your platform.
$ sudo pipinstall /tmp/tensorflow_pkg/tensorflow-1.0.0-py2-none-any.whl
 
安装完成之后，进入python测试，
$ python
...
>>> import tensorflow as tf
>>> hello = tf.constant('Hello, TensorFlow!')
>>> sess =tf.session()
>>> PRint(sess.run(hello))
Hello,TensorFlow!
>>> a = tf.constant(10)
>>> b = tf.constant(32)
>>> print(sess.run(a + b))
42
>>> 
如果出现tensorflowfailed call to cuinit
sudoapt-get install nvidia-modprobe