This is a installation guide for Windows (tested only on Windows 10) using pip
-
Get Python 3.4-3.7 64-bit only
-
Add to PATH
-
Get pip (comes with Python installation)
-
(Optional)
pip install virtualenv
-
pip install --upgrade tensorflow
-
Test with
python -c "import tensorflow as tf;print(tf.reduce_sum(tf.random.normal([1000, 1000])))"
If you get a lot of not located warnings but get
tf.Tensor(-1565.4648, shape=(), dtype=float32)
Then it is a success
-
(Optional for Nvidia Graphics cards GT 430 or GTX 465 or 410M or higher;
See full list @ link)
-
The latest NVIDIA® GPU drivers
-
The CUDA® Toolkit
-
The cuDNN SDK (Requires a Nvidia Account)
-
Add to PATH the following (Note, must be v10.1 as of 1/12/20, otherwise check the documentation in the link here)
-
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\bin
-
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\extras\CUPTI\libx64
-
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.1\include
-
Where ever you unziped the cuDNN SDK's bin directory. i.e.
C:\tools\cuda\bin
-
-
-
Repeat step 6 but you should see a bunch of successes and
tf.Tensor(-1565.4648, shape=(), dtype=float32)
Post an Issue to the Issues tab in GitHub
- unspecific documentation: TensorFlow