TensorFlow dependencies required. How to run TensorFlow on Windows

Question

I am interested in running TensorFlow on Windows, but currently I understand that this is not possible due to some of the dependencies not used on Windows, such as Basel.

The need arises because, as I understand it, the only way to access the GPU from TensorFlow is not a virtual Linux installation. I really understand that I can double boot on Linux, but prefer to avoid this route.

To solve the problem, I need a whole chain of dependencies to create TensorFlow, as it was interesting if it already existed.

I also understand that I can capture the output of the assembly when building from the source code as a solid start, but I would like to avoid this work if it is already known.

There is a beta version of Bazel that runs on Windows - https://github.com/dslomov/bazel-windows

See the related GitHub issue for running TensorFlow on Windows. - https://github.com/tensorflow/tensorflow/issues/17

Another reason to run on Windows is the ability to connect to the Xbox One.

I found a possible answer, still need to check it. This will create a dependency graph as a point file.

$ bazel query 'deps(//tensorflow/tools/pip_package:build_pip_package)' --output graph > tensorflow.dependency.dot 

Answer 1

There are currently three main options for building and / or running TensorFlow on Windows:

• You can install the TensorFlow 0.12rc0 GPU-enabled PIP package from PyPI: pip install tensorflow-gpu
• You can create a GPU-enabled PIP package yourself using the CMake experimental build . It also gives you the ability to work with TensorFlow in Visual Studio. The documentation for this assembly can be found here .
• There is preliminary support for creating TensorFlow using Bazel for Windows . However, we are still smoothing out some bugs with this assembly.

Answer 2

It may not be exactly what you want one of the ways to run TensorFlow on Windows is to install a virtual machine (VMWare player v12 can be used for non-commercial use), and then install Ubuntu in this and finally TensorFlow in Ubuntu. Works well for me.

Answer 3

Since the beginning of 2017, Tensorflow is now officially supported on Windows and can be installed via pip :

 pip install --upgrade tensorflow pip install --upgrade tensorflow-gpu 

or by fetching packages directly (choose the one that suits your needs, for example x64 / gpu)

 # x86 / CPU pip install --upgrade https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-1.0.0-cp35-cp35m-win_x86_64.whl # x64 / CPU pip install --upgrade https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-1.0.0-cp35-cp35m-win_amd64.whl # x64 / GPU pip install --upgrade https://storage.googleapis.com/tensorflow/windows/gpu/tensorflow_gpu-1.0.0-cp35-cp35m-win_amd64.whl