How to install Tensorflow 2.5 and CUDA 11.2 with a CPU and a GPU on Windows 10

#tensorflow #deeplearning #cuda #gpu #rtx30 #rtx3060 #rtx3070 #rtx3080 #rtx3090 #amd

How to install Tensorflow 2.5, CUDA 11.2.1, and CuDNN 8.1, for Windows 10, with full deep learning support for an Nvidia GPU RTX 30 series card, like the Nvidia RTX 3060, RTX 3060TI, RTX 3070, RTX 3080, and RTX 3090. Since CUDA is backward compatible it should also work for RTX 20 series cards or older.

But before I install CUDA, I will do some benchmarking of Tensorflow 2.5 without a GPU. I will be training an image classification model that I developed in a previous video, with my AMD Ryzen 5900X CPU.

The AMD Ryzen 5900X CPU has 12 cores and 24 threads, and it is a decent CPU all around, perfect for a beginner in Deep Learning like myself.

Once I complete the CUDA setup, I will try again to complete the training of the same model with my Nvidia RTX 3070 GPU. It has 5888 CUDA cores. Can it beat the AMD Ryzen 5900X CPU? It definitely should.

✅ Recommended Courses
(Includes Amazon Affiliate Links)
==================================

✅ Recommended Playlists
==================================

✅ My Favorite Tech
(Includes Amazon Affiliate Links)
==================================

✅ My RTX 3070 and Ryzen 5900 Machine Learning PC Build
(Includes Amazon Affiliate Links)
=======================================================

As an Amazon Associate, I earn a small commission from qualifying purchases on the amazon links above. It costs you nothing but helps me in keeping the content coming.

✅ ARTICLE

✅ BUSINESS ENQUIRIES

✅ SUPPORT ME

✅ TIMESTAMPS

0:00 Introduction
2:15 Installing Anaconda
3:18 Installing latest Nvidia Game-Ready graphics driver
3:49 Creating new Anaconda environment for Tensorflow 2.5
4:30 Installing Tensorflow 2.5 rc 1 with pip, matplotlib and pandas
6:00 Installing JupyterLab from Anaconda
6:45 Testing Tensorflow model training works with my AMD Ryzen 5900X CPU
10:00 Reviewing training times for Tensorflow with the Ryzen 5900X CPU
11:40 Installing CUDA 11.2.1
12:08 Installing cuDNN 8.1
13:00 Copying cuDNN files to CUDA folder
15:00 Setting CUDA environment variable
16:00 Test Tensorflow model training with my RTX 3070 GPU
17:49 Compare GPU vs CPU for Deep Learning

[dalilabdouraman]

Thanks a lot Code mental.I spent almost 3 days and nights looking on how to use my GTX 1050 with tensorflow...all the explanations were easy to understand..thx a lot

[w.abdalaziz]

Finally it is working .... Thanks a lot mate.. :)

[kannanv]

Your earlier Video was a lot of help in installing my Cuda 11.3, but due to bugs and other coding issues in my TF2 OD API, I am still unable to run training instances using my GE RTX 3070 Nvidia gave up on Windows, and switched to Ubuntu 18.04 on a dual boot in my new I will again switch back to Windows 10 and follow the steps in your video and see, if I am able to succeed for all these Videos, and really

[budzn]

I'm having trouble downloading the *dataset.csv files:

FileNotFoundError: [Errno 2] No such file or directory: 'nativity_dataset.csv'

Can anyone help? Thanks for the Tutorial!

[magdalenanedic]

Thanks for the explanation in the video! I am interested if this process of installation works the same for NVIDIA GeForce RTX 3060 Laptop GPU?

[blackbigdeath]

Thanks, its works. But i install directly without conda.

[aliali-sjjp]

Thank you for this wonderful explanation . I have a question I don't know what made me think about this . but I have a new laptop with a processor intel(R) Core(TM) i7-7500U CPU @ 2.70GHz 2.90 GHz. Can I Set Up CUDA, CUDNN, and TensorFlow GPU on Windows 10>>>>thanks very much

[tanhungtruong]

excuse me, can it works with NVIDIA Geforce RTX 1650 laptop GPU?

[xjiyhzc]

I have a problem want to ask. I'm failed in installing CUDA, I download the 11.2 version . In the first step system check, it told me cannot continue because of I have already have a new version of the NVIDIA Frameview SDK installed. How can I solve it, I try to uninstalled and reload. (My laptop is hp pavilion gaming laptop 15, GTX1650)

[aashish]

After installing CUDA11.2.1, it is not showing me those folders - bin, lib etc. Only extras folder is available. Do you know what could be wrong?

[kritikseth]

Can I Install Cuda 11.4 instead of 11.2?