[FIXED] Tensorflow on GPU WIndows 11

Issue

I have Windows 11, nVidia GTX 1650 Ti and I am trying to make Tensorflow work on the GPU. And one absolute condition: I don't want to even think about Linux. It has to work on Windows. Currently on Python 3.11.2, Tensorflow 2.13.1

I tried official nVidia guide to install cuda and cudnn, some articles from medium.com and still nothing. I switched from PyCharm to Visual studio Code. Nothing. Installing through conda. Nothing. I tried newest cuda and cudnn versions, cuda 11.8 with cudnn 8.6.0 or with cudnn 8.9.7 and still nothing. Tensorflow just can't see my GPU. And yes, nVidia Game drivers are updated. I have no other ideas except linux, but i would rather jump from the window than use linux. I'm a mathematician, not IT specialist, so please give me step-by-step solution. Thanks in advance

import sys
import keras
import tensorflow as tf
import numpy as np

print(f"Tensor Flow Version: {tf.__version__}")
print(f"Keras Version: {keras.__version__}")
print()
print(f"Python {sys.version}")
gpu = len(tf.config.list_physical_devices('GPU'))>0
print("GPU is", "available" if gpu else "NOT AVAILABLE")

result of code above

Solution

You need to downgrade your TF version to 2.10 or lesser to work on native-Windows (not recommended though)

GPU support on native-Windows is only available for 2.10 or earlier versions, starting in TF 2.11, CUDA build is not supported for Windows. The correct supported CUDA and cuDNN version for Tensorflow 2.10 are CUDA 11.2 and cuDNN 8.1 as mentioned in this tested build config

link : https://discuss.tensorflow.org/t/issues-installing-tensorflow-on-windows-11-with-nvidia-2060/19663

Caution: TensorFlow 2.10 was the last TensorFlow release that supported GPU on native-Windows. Starting with TensorFlow 2.11, you will need to install TensorFlow in WSL2, or install tensorflow or tensorflow-cpu and, optionally, try the TensorFlow-DirectML-Plugin

link: https://www.tensorflow.org/install/pip#windows-native

Answered By - Goku - stands with Palestine