CUDA not Enabled in 4060 - Any guide in how to use GPU in Deep Learning?

AI & Data Science Deep Learning (Training & Inference) cuDNN
1

Hi, This is Pau!

I’m working with a computer with a dedicated NVIDIA Geforce RTX 4060 GPU, and I just realized, with the help of customer support, that is not CUDA-Enabled.

The problem here is that I’m working with deep learning, creating models with large training data sets and training parameters. My question is if anyone knows how to use my GPU to help preprocess the data and making the model go faster.

I tried installing WSL v2, with Ubuntu, installed Cuda Toolkit and cuDNN, all this before knowing that is not CUDA Enabled, to work with Tensorflow, Keras and Pytorch along with others libraries.

Well, If anyone knows anything, help would be great!

Thanks!

1 Like
2

Did you find any solution?

3

HI Stev!

I did not. I found that you can have a docker with all the Nvidia Drivers and Jupyter Notebook integrated in order to perform this kind of operations but I didn’t notice a major improvement in the performance or time response of the models.

In case you are in the same spot that I was few months ago, I recommend you to use cloud services as AWS, Azure or Google Cloud platform. They seem to be great for this purpose