This Op computes the double of an input tensor (2 * input)
Tensorflow 1.11 built from source with GPU support.
Ubuntu 16.04.
Place a BUILD file with the .cc files in the tensorflow/core/user_ops directory.
Run the following command in the tensorflow directory to compile double_out.so
bazel build --config opt //tensorflow/core/user_ops:double_out.so
The jupyter file (.ipynb) shows how to use the op in Python.