Step-by-Step Guide: Install NVIDIA GPU Driver, CUDA Toolkit, cuDNN, and TensorRT on Windows 11

🚀 Welcome to our comprehensive tutorial on installing and configuring NVIDIA graphics card drivers, CUDA Toolkit, cuDNN, and TensorRT on Windows 11! If you’re looking to boost your machine learning and AI capabilities, this video is for you. Follow along as we break down each step, ensuring you get your system up and running smoothly.
In This Video, You’ll Learn:
NVIDIA Graphics Card Driver Installation: How to download and install the latest drivers for your NVIDIA GPU.
CUDA Toolkit Installation: Step-by-step instructions to install CUDA Toolkit, enabling GPU acceleration for your projects.
cuDNN Installation: How to integrate NVIDIA's deep neural network library for high-performance neural network training.
TensorRT Installation: Optimize and deploy your AI models efficiently with NVIDIA TensorRT.
Chapters:
0:00 Introduction
0:43 Installing NVIDIA Graphics Card Driver
3:11 Prerequisites
3:46 Visual Studio Community Edition
4:55 Nvidia Nsight Visual Studio
5:57 Installing Mini conda
8:45 Cuda Toolkit 12.4.1
12:12 cuDNN Configuration
14:55 TensorRT installation
20:35 Pytorch with CUDA
23:05 Conclusion
Useful Links:
- NVIDIA App: [NVIDIA's official software page](www.nvidia.com...)
- Visual Studio: [Visual Studio Community Edition](visualstudio.m...)
- Nvidia Nsight VS: [Nvidia Nsight for Visual Studio](developer.nvid...)
- Miniconda: [Miniconda installer links](docs.anaconda....)
- CUDA Toolkit: [CUDA Toolkit Archive](developer.nvid...)
- cuDNN: [cuDNN Archive](developer.nvid...)
- TensorRT: [TensorRT Download](developer.nvid...)
- TensorRT Documentation: [TensorRT Installation Guide](docs.nvidia.co...)
Join The Community:
Subscribe to our channel for more tutorials on AI, machine learning, and tech tips. Hit the notification bell so you never miss an update!
Hashtags:
#NVIDIA #CUDA #cuDNN #TensorRT #Windows11 #MachineLearning #AIGuide #TechTutorial

[@AdityalikeThe]

Best cuda installation guide out there.

[@ToonMixYt]

Thanks bro

[@Piyush.A]

Thank you! Much appreciated!

[@nandyaladineshmurali4505]

Hello there, I have the CUDA 12.5 installed on my system(Windows 11). And trying to install the TensorRT - 10.2.0.19 which match my cuda version. Everything is fine until 20:00 . But there is no onnx_graphsurgeon folder in my directory. I tried reinstalling the tensorRT multiple times but no use. Please help me out.

[@see_beautiful_peaceful_world]

Thank you very much

[@michaeldang4288]

Thank you so much, this is the exact video i've been searching for. Your video moves at the perfect pace and everything is easy to follow. will definitley be watching your other videos after this one. great work!

[@TechJotters24]

Thank you @michaeldang4288

[@keanureecvers5589]

awesome video , i followed you step by step and it installed successfully.Lmao after 2 months of struggling your video helped alot🙏

[@TechJotters24]

You are welcome. I am happy it works for you.

[@KarmaRayshar]

cuda binary kernel for this graphics card compute capability (8.9) not found, I can't render animation in blender. Any Idea? Thanks

[@TEXASMOFO]

Very informational video... clear and concise. 👍

[@TechJotters24]

Thank you!!!

[@LikquidDutch]

I dont have graphsurgeon in my tensor zip file im using cuda 11.8 is it required ?

[@nihar1409]

Does this work with jupyterlab? In powershell it showed True but in jupyterlab it shows false? I am using same environment, but Cuda 12.5 with pytorch installation link for 12.4 is this the reason cud.is_available() is returning False in jupyterlab?

[@TechJotters24]

Hi, Tensorflow 2.16.1 only support Cuda Toolkit 12.3. I also found this issue, powershell shows GPU available but Jupyter not, and basically tensorflow don't get GPU support.

[@Guto7Schmitt]

There is no onnx_graphsurgeon file

[@mohamadelsawi]

hello my friend i did everything you said , and thank you alot for your effort , i have question when i reached to the part you will install pytorch , i found that i have it on my pc , should i do the same as you did , and why to do that :)
another question ^^
will i open conda to be in env everytime i restart my pc or it will be permanent work , as i will use stable diffusion with TensorRt extension

[@TechJotters24]

Hi, I installed PyTorch to test whether it can you the GPU or not. If you don’t need PyTorch, you can skip it. It’s only for testing.

[@marcosvazquez1648]

very nice job

[@TechJotters24]

Thank you

[@yuval_kahan]

why you didn't install pytorch 2.3.0 ?

[@TechJotters24]

Hi, I installed Cuda toolkit 12.4 and only the pytorch nightly version supports Cuda Toolkit 12.4 right now. If you want to install PyTorch 2.3.0, It; better to install Cuda toolkit 12.1. The installation process is same.