linux平台下编译 TensorFlow 的 C/C++ 接口

安装环境：
Ubuntu 16.04
Python 3.5
CUDA 9.0
cuDNN 7
Bazel 0.17.2
TensorFlow 1.11.0

1. 安装 Bazel

• 安装 JDK sudo apt-get install openjdk-8-jdk

• 添加 Bazel 软件源

• 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 -

• 安装并更新 Bazel sudo apt-get update && sudo apt-get install bazel

•  

• 2. 编译 TensorFlow 库

• TensorFlow 源码：https://github.com/tensorflow/tensorflow

• 进入源码根目录，运行 ./configure 进行配置。可参考 官网 -> Build from source -> View sample configuration session 设置，主要是 Python 的路径、CUDA 和 CUDNN 的版本和路径以及显卡的计算能力 

• 进入 tensorflow 目录进行编译，编译成功后，在 /bazel-bin/tensorflow 目录下会出现 libtensorflow_cc.so 文件

• C版本： bazel build :libtensorflow.so
  C++版本： bazel build :libtensorflow_cc.so
  

• 3. 编译其他依赖
  进入 tensorflow/contrib/makefile 目录下，运行./build_all_linux.sh，成功后会出现一个gen文件夹

  若出现如下错误 /autogen.sh: 4: autoreconf: not found    ，安装相应依赖即可    sudo apt-get install autoconf automake libtool

  4. 测试

• Cmaklist.txt

• cmake_minimum_required(VERSION 3.8)
  project(Tensorflow_test)

  set(CMAKE_CXX_STANDARD 11)

  set(SOURCE_FILES main.cpp)

  
  include_directories(
          /media/lab/data/yongsen/tensorflow-master
          /media/lab/data/yongsen/tensorflow-master/tensorflow/bazel-genfiles
          /media/lab/data/yongsen/tensorflow-master/tensorflow/contrib/makefile/gen/protobuf/include
          /media/lab/data/yongsen/tensorflow-master/tensorflow/contrib/makefile/gen/host_obj
          /media/lab/data/yongsen/tensorflow-master/tensorflow/contrib/makefile/gen/proto
          /media/lab/data/yongsen/tensorflow-master/tensorflow/contrib/makefile/downloads/nsync/public
          /media/lab/data/yongsen/tensorflow-master/tensorflow/contrib/makefile/downloads/eigen
          /media/lab/data/yongsen/tensorflow-master/bazel-out/local_linux-py3-opt/genfiles
          /media/lab/data/yongsen/tensorflow-master/tensorflow/contrib/makefile/downloads/absl
  )

  add_executable(Tensorflow_test ${SOURCE_FILES})

  target_link_libraries(Tensorflow_test
          /media/lab/data/yongsen/tensorflow-master/bazel-bin/tensorflow/libtensorflow_cc.so
          /media/lab/data/yongsen/tensorflow-master/bazel-bin/tensorflow/libtensorflow_framework.so
          )

• 创建回话

• #include <tensorflow/core/platform/env.h>
  #include <tensorflow/core/public/session.h>
  #include <iostream>

  using namespace std;
  using namespace tensorflow;

  int main()
  {
      Session* session;
      Status status = NewSession(SessionOptions(), &session);
      if (!status.ok()) {
          cout << status.ToString() << "\n";
          return 1;
      }
      cout << "Session successfully created.\n";
      return 0;
  }

• 查看 TensorFlow 版本

• #include <iostream>
  #include <tensorflow/c/c_api.h>

  int main() {
     std:: cout << "Hello from TensorFlow C library version" << TF_Version();
      return 0;
  }

  // Hello from TensorFlow C library version1.11.0-rc

• 若提示缺少某些头文件则在 tensorflow 根目录下搜索具体路径，然后添加到 Cmakelist 里面即可。

该内容借鉴：https://blog.csdn.net/seniusen/article/details/82972335