Ubuntu 环境 TensorFlow 源码编译安装
基于（Ubuntu 14.04LTS/Ubuntu 16.04LTS/）
一、编译环境
1) 安装 pip
sudo apt-get install python-pip python-dev
2）安装JDK 8
sudo apt-get install openjdk-8-jdk
Ubuntu 14.04 LTS 还需要：
sudo add-apt-repository ppa:webupd8team/java
sudo apt-get update && sudo apt-get install oracle-java8-installer
3）安装Bazel
A: 添加 Bazel URI 到 package source
echo "deb [arch=amd64] http://storage.googleapis.com/bazel-apt stable jdk1.8" | sudo tee /etc/apt/sources.list.d/bazel.list
curl https://bazel.build/bazel-release.pub.gpg | sudo apt-key add -
B:更新&安装
sudo apt-get update
sudo apt-get install bazel
如果已经安装过，更新则：
sudo apt-get upgrade bazel
C:设置环境变量
一次执行
export PATH="$PATH:$HOME/bin"
直接添加到.bashrc ，打开bashrc 最后一行加入(PATH="$PATH:$HOME/bin")
vim ~/.bashrc
PATH="$PATH:$HOME/bin"
4)安装其他依赖包
sudo apt-get install libcupti-dev
sudo pip install --upgrade protobuf
sudo apt-get install git python-dev python3-dev python-numpy python3-numpy python-six python3-six build-essential python-pip python3-pip python-virtualenv swig python-wheel python3-wheel libcurl3-dev libcupti-dev
apt-get install libglib2.0-dev zlib1g-dev
sudo apt-get install librdmacm-dev
5) 如果要GPU支持需要
A:安装/更新GPU驱动
sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt update
B:Nvidia Toolkit 8.0 & CudNN
在https://developer.nvidia.com/cuda-toolkit下载对应的版本
sudo sh cuda_8.0.61_375.26_linux.run --override --silent --toolkit
会将cuda安装到: /usr/local/cuda
C:安装CudNN
在 https://developer.nvidia.com/cudnn 下载对应的版本
解压到 /usr/local/cuda
tar -xzvf cudnn-8.0-linux-x64-v6.0.tgz
sudo cp cuda/include/cudnn.h /usr/local/cuda/include
sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn*
D: 配置环境变量
~/.bashrc 添加
export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/usr/local/cuda/lib64:/usr/local/cuda/extras/CUPTI/lib64"
export CUDA_HOME=/usr/local/cuda
然后使环境变量生效
source ~/.bashrc
二、 TensorFlow 源码下载、编译、安装
1）下载tensorflow 源码
git clone https://github.com/tensorflow/tensorflow
2）配置TensorFlow
到TensorFlow的根目录执行
./configure
注：出于国情原因下面的一定选N
Do you wish to build TensorFlow with Google Cloud Platform support? [y/N]
Do you wish to build TensorFlow with Amazon S3 File System support? [Y/n]
Do you wish to build TensorFlow with Hadoop File System support? [y/N]
3）编译安装
bazel编译pip 的安装包，然后通过 pip 安装
1) bazel编译
bazel build -c opt //tensorflow/tools/pip_package:build_pip_package
2) 生成安装包
bazel-bin/tensorflow/tools/pip_package/build_pip_package /tmp/tensorflow_pkg
2017年 12月 12日 星期二 13:32:22 CST : === Output wheel file is in: /tmp/tensorflow_pkg
3) 安装
sudo pip install /tmp/tensorflow_pkg/tensorflow-1.4.0-cp27-cp27mu-linux_x86_64.whl
注意： 2）生成安装包的目录，tensorflow-1.4.0-cp27-cp27mu-linux_x86_64.whl在=== Output 提示的 /tmp/tensorflow_pkg下
安装过程会下载一些依赖的包和库，最后成功提示：
Successfully installed absl-py-0.1.6 backports.weakref-1.0.post1 bleach-1.5.0 enum34-1.1.6 funcsigs-1.0.2 html5lib-0.9999999 markdown-2.6.10 mock-2.0.0 numpy-1.13.3 pbr-3.1.1 tensorf
三、遇到问题
编译时出现如下错误：
ERROR: /home/duanyufei/source/TensorFlow/tensorflow/tensorflow/contrib/gdr/BUILD:52:1: C++ compilation of rule '//tensorflow/contrib/gdr:gdr_memory_manager' failed (Exit 1)
tensorflow/contrib/gdr/gdr_memory_manager.cc:28:27: fatal error: rdma/rdma_cma.h: No such file or directory
compilation terminated.
Target //tensorflow/tools/pip_package:build_pip_package failed to build
Use --verbose_failures to see the command lines of failed build steps.
INFO: Elapsed time: 323.279s, Critical Path: 33.69s
FAILED: Build did NOT complete successfully
解决办法
sudo apt-get install librdmacm-dev
四、测试 hello word！
在终端打开python，运行如下代码
>>> import tensorflow as tf
>>> hello = tf.constant('Hello, TensorFlow!')
>>> sess = tf.Session()
>>> print(sess.run(hello))
结果：
Hello, TensorFlow!