Installing TensorFlow GPU on Windows | Latest Way | ALL Error Fixed !!

Hello World, Welcome to our channel where we talk about tech and programming and Data Science stuffs.
Installing TensorFlow on GPU is not an easy process so today I will be showing you the right way how to install the TensorFlow GPU version on your Windows machine.
Links:

Installation Process:
1. Install VC_redist.x64
2. CUDA
3. Copy files cuDNN
4. Install TensorFlow

🤗🤗
🤗🤗
Social Media:

======= TAGS ================
python,windows,keras,tensorflow,anaconda,miniconda,install tensorflow,neural network,deep learning,sklearn,pandas,jupyter,conda,conda environment,artificial intelligence,data science,jupyter notebooks,wustl,washington university,gpu,nvidia

[phatcao]

it took me a lot of time .... thanks very much!!!

[aminem]

Big thanks man !
I tried days to avoid uninstalling everything, but finally I got to admit that I have no choice except following your advice, and things worked like a charm :)
Thank you :)

[spartacuspolok]

Thanks...
It really helped me a lot...
Finally, , , I could install it properly...

[CyberCreed]

Please follow this step of the installation process
Installation Process:
1. Install VC_redist.x64
2. CUDA
3. Copy files cuDNN
4. Install TensorFlow
And if you want to work with Keras then you might get an error like this "Keras requires TensorFlow 2.2 '
So to fix it just install tensorflow-gpu==2.2.0
Command :
pip install tensorflow-gpu==2.2.0

For any query join our telegram group

[ghostplays]

Bro i got jit compilation error while running on gpu but it runs in cpu what is that

[ravinash]

you are my hero man!!!!

to verify if this works select cuda under performance tab in task manager and see utilization rate.
for people who can't see cuda in performance tab of task manager, turn off 'hardware-accelerated GPU scheduling' in graphics settings in
control panel.

[aritraray]

will this work for any cuda enabled gpu irrespective of the compute capability version of cuda correaponding to my gpu??...which is listed in

[RitwikDandriyal]

Hey. Is it safe to uninstall all nvidia drivers at start from apps and features? Including stuff like nvidia control panel and stuff like that?

[AlexRyzhickov]

Whtat is your Python (Anaconda) version ? Is your version younger than 3.8 ?

[randomforrest]

weird. It definitely works in jupyter notebooks (training my cnn took 53 seconds per epoch, now it takes 12 seconds in the first and 8 seconds in the other epochs)..but using spyder with the same environment throws an cuDNN error... hmm

thanks for your tutorial

[atharvameshram]

Do I have to remove Nvidia folder from program files (x86)

[IItianrider]

what about the MSVC and bazel installation?

[mahalim.]

Hello Brother.
I am recently using NVIDIA GPU. I am learning Adobe premiere pro CC. But my problem is when go to Export my video and unable to export Due to shows Error Compelling Movie/GPU Render Error. Please advice the solution. Whereas I have downloaded Cuda 460.89. and NVIDIA package 452.06.
I am from Dhaka

[hrishikeshbhor]

Removing NVIDIA folders will affect on gameing performance....?

[spyrostsatsis]

Do i need vs studio in order to use cuda for ML?

[prabhavsingh]

Hi!
Thanks for the video! I went through all the steps and I am pretty sure I did it correctly but it still shows 0 GPU's available.
I checked the command prompt and it showed this error - failed call to cuInit: CUDA_ERROR_UNKNOWN: unknown error
Any tips?

[thanhphamchi]

OMG You saved my life (y). Thank you!!!

[dipayandeb]

getting UnknownError: Fail to find the dnn implementation. [Op:CudnnRNN]

[MaicoTriaca]

Why i need to delete all the nvidia folders?

[akshhay]

Is anaconda necessary, cant we use pip?