The environment that
Hardware:
CPU: E5-2673 V3
Main board: Asus Z10Z10PA-D8
Memory: Samsung DDR4 2400 ECC-R 16G*2
Graphics card: MSI GTX1080Ti AERO
Solid state drive: Samsung PM961 256G M.2
Mechanical hard disk: Seagate 2T
Power supply: Great Wall 1250W
Case: ANTEC P8
Fan: 10 heat pipes are silent
System: Windows 10 Enterprise Version
The early stage of the work
Latest version of NVIDIA driver
Visual Studio 2015 Visual Studio Community 2015
Note: After testing, Visual Studio 2013 also works.
Install CUDA and cuDNN
Note: At the time of this article, the latest versions of CUDA and cuDNN are not supported by TensorFlow, so this article uses the following two versions:
1. CUDA 8.0
Download address: developer.nvidia.com/cuda-downlo…
Install it directly. You can also go to my Baidu cloud download: link: pan.baidu.com/s/1dFHD37F…
2. CuDNN v6.0
Download address: developer.nvidia.com/rdp/cudnn-d…
You can also go to my Baidu cloud download: link: pan.baidu.com/s/1gftmqR5…
Unzip the package and overwrite it to C: Program Files NVIDIA GPU Computing Toolkit CUDA V8.0.
3. Install Tensorflow GPU 1.4
Because Anaconda provides a complete library of scientific computing, you can use Anaconda directly for the installation.
4. Installation Anaconda
Download address: www.anaconda.com/download/
Download Anaconda 5 with Python 3.6 64bit and install it directly.
5. Install TensorFlow GPU 1.4 on Anaconda
PIP install –ignore-installed –upgrade tensorflow-gpu standby Other commands:
Deactivate tf # deactivate tf # conda remove –name tf –all
The following components will be automatically installed
numpy
wheel
tensorflow-tensorboard
six
protobuf
html5lib
markdown
werkzeug
bleach
setuptools
Use the following code to test the installation
6. Test.
On the cli:
activate tf
python
Enter the following code:
import tensorflow as tf hello = tf.constant(‘Hello, TensorFlow! ‘) sess = tf.Session() print(sess.run(hello))
No error is configured.
Editing in the 2018-01-21