ubuntu-14.04安装最新tensorflow记录

1.安装英伟达驱动

./NVIDIA-Linux-x86_64-384.69.run

nvidia-smi成功表示驱动ok

2.安装cuda

dpkg -i cuda-repo-ubuntu1404-8-0-local-ga2_8.0.61-1_amd64.deb

apt-get update

apt-get install cuda

安装patch2（也可以不装）dpkg -i cuda-repo-ubuntu1404-8-0-local-cublas-performance-update_8.0.61-1_amd64.deb

3.降低gcc版本到5.0以下（ubuntu-14不需要，因为已经是gcc-4.8.4，ubuntu-16才要）

sudo apt-get install g++-4.9

sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-4.9 20

sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-5 10

sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-4.9 20

sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-5 10

sudo update-alternatives --install /usr/bin/cc cc /usr/bin/gcc 30

sudo update-alternatives --set cc /usr/bin/gcc

sudo update-alternatives --install /usr/bin/c++ c++ /usr/bin/g++ 30

sudo update-alternatives --set c++ /usr/bin/g++

4.安装cudnn

tar zxf cudnn-8.0-linux-x64-v6.0.solitairetheme8

把cuda下的文件拷贝到/usr/local/cuda

5.python3及tensorflow

apt-get remove python2.7

apt-get install python3.5

cd /usr/bin

ln -s python3.5 python

一定要确保运行python版本是3.5.x

wget https://bootstrap.pypa.io/get-pip.py

python get-pip.py

sudo pip3 install setuptools --upgrade

sudo pip3 install ipython

cp /usr/local/bin/pip3.5 /usr/bin/pip3

6.安装tensorflow

pip3 install tensorflow-gpu==1.2

说明：tensorflow-1.2是2017.6发布的版本，应该足够

7.测试：

$ python

>>> import tensorflow as tf

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

>>> sess = tf.Session()

>>> sess.run(hello)

'Hello, TensorFlow!'

>>> a = tf.constant(10)

>>> b = tf.constant(32)

>>> sess.run(a + b)

42

>>>