self_check | self_check

Here are some instructions: https://www.google.com/amp/s/mc.ai/how-to-install-cuda-10-and-cudnn-for-tensorflow-gpu-on-windows-10/amp/ Just download the dll, extract it to your prefered path and add this path to your environment variables.

tensorflow doesn't support python 3.7 version. Requires Python 3.4, 3.5, or 3.6

see below https://www.tensorflow.org/install/pip

Here is what I did.

Step 1) Installed 'NVIDIA GEFORCE EXPERIENCE' in my computer to check my Driver version.

Step 2) The driver version was an old one. Update was available. So I updated my Graphic driver.

My GPU properties now are:-

there are two module of tensorlfow:'tensorflow','tensorflow-gpu' on cpu you need to install tensorlfow with pip install tensorflow or on conda conda install tensorflow

EDIT for second question:

If a TensorFlow operation is placed on the GPU, then the execution engine must have the GPU implementation of that operation, known as the kernel.
If the kernel is not present, then the placement results in a runtime error. Also, if the requested GPU device does not exist, then a runtime error is raised.
The best way to handle is to allow the operation to be placed on the CPU if requesting the GPU device results in an error.

One answer would be to remove all GPU configs and second would be soft placement if GPU is not found as explained above use config.allow_soft_placement = True

To run tensorflow with GPU support you'll need a NVIDIA graphic card and the CUDA libraries installed (see requirements here), unfortunately your Intel(R) HD Graphics 520 won't do.

You can settle for the CPU version (pip3 install --upgrade tensorflow), which is still good to start learning.

UPDATE: another solution is to use the online service Google Colab, which allows you to borrow a GPU from Google's cloud infrastructure for learning purposes.

I had a similar problem, and had to be very careful about the version of CUDA, and the version of CuDNN. I hit the exact error you are hitting, and fixed it by going through what I documented here: http://www.laurencemoroney.com/installing-tensorflow-with-gpu-on-windows-10/

Give it a try! :)

(The most common failure I've found is that you download the latest CUDA, and not the matching CUDA. Right now CUDA is at 9.1, but TF requires 9.0 -- your error says cudart90.dll) -- so find the 9.0 drivers, download and install them. Then run TensorFlow. It might then fail on the CuDNN drivers, which is good, becuase you know CUDA is right. Then download the right CuDNN drivers (matching the version # in the error) and try again.)

You should also have cudnn on your computer as the error code says. Please see: https://developer.nvidia.com/cudnn

Problem solved. I reinstalled CUDA V9.0 and cuDNN with 7.0.4 and then reinstalled tensorflow-gpu with pip uninstall tensorflow-gpu pip install tensorflow-gpu

It looks like there was some other weird stuff going on. The biggest thing I noticed that was odd is that sudo pip --version gave a different version than just pip --version.

Doing pip install --upgrade https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.8.0-cp35-cp35m-linux_x86_64.whl upgraded and allowed me to import tensorflow. (Note that doing sudo pip did not work.)

Thank you @HarisNadeem for the solution and for helping diagnose the issue!