机器学习：使用 NVIDIA JetsonTX2 - 安装TensorFlow

安装 TensorFlow

1. 安装依赖套件

    $ sudo apt-get install default-jdk libcupti-dev
    $ export JAVA_HOME='/usr/lib/jvm/java-8-openjdk-arm64/'

2. 取得 TensorFlow 编译脚本

    $ git clone git://github.com/jetsonhacks/installTensorFlowTX2
    $ cd installTensorFlowTX2

3. 执行编译脚本

    $ ./installPrerequisitesPy3.sh
    $ ./cloneTensorFlow.sh
    $ ./setTensorFlowEVPy3.sh
    $ ./buildTensorFlow.sh
    $ ./packageTensorFlow.sh

4. 安装 TensorFlow

   1. 建立虚拟开发环境

       $ virtualenv TensorFlow

   2. 进入虚拟开发环境

       $ cd TensorFlow
       $ source bin/active

   3. 安装 TensorFlow 到虚拟环境

       pip3 install $HOME/<TensorFlow 的 .whl 安装封包>

   4. 测试 TensorFlow

      1. Hello World

          import tensorflow as tf
          hello = tf.constant('Hello, TensorFlow on NVIDIA Jetson TX2!')
          sess = tf.Session()
          print(sess.run(hello))

         输出

          Hello, TensorFlow on NVIDIA Jetson TX2!

      2. 运算单元

          import tensorflow as tf
          # Creates a graph.
          a = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[2, 3], name='a')
          b = tf.constant([1.0, 2.0, 3.0, 4.0, 5.0, 6.0], shape=[3, 2], name='b')
          c = tf.matmul(a, b)
          # Creates a session with log_device_placement set to True.
          sess = tf.Session(config=tf.ConfigProto(log_device_placement=True))
          # Runs the op.
          print(sess.run(c))

         输出

          name: NVIDIA Tegra X2
          major: 6 minor: 2 memoryClockRate (GHz) 1.3005
          MatMul: (MatMul): /job:localhost/replica:0/task:0/gpu:0
          b: (Const): /job:localhost/replica:0/task:0/gpu:0
          a: (Const): /job:localhost/replica:0/task:0/gpu:0
          [[ 22.  28.]
          [ 49.  64.]]