Check compatible python version in wheel file list at the end of the page:
https://www.tensorflow.org/install/pip?lang=python3
conda create -p D:\Code\CondaEnvs\tf-gpu python=3.6
install pip:
conda install pip
https://www.tensorflow.org/install/pip?lang=python3
example:
pip install https://storage.googleapis.com/tensorflow/windows/gpu/tensorflow_gpu-2.0.0-cp36-cp36m-win_amd64.whl
Compatible versions list at the end of the page:
https://www.tensorflow.org/install/source_windows
Download:
https://developer.nvidia.com/cuda-toolkit-archive
Select LATEST version, the version mentioned in Tensorflow page is not necessarily the correct one, you can still get an error like:
tensorflow/stream_executor/cuda/cuda_dnn.cc:319] Loaded runtime CuDNN library: 7.4.2 but source was compiled with: 7.6.0. CuDNN library major and minor version needs to match or have higher minor version in case of CuDNN 7.0 or later version.