DEV Community

Cover image for How to Set Up Python, CUDA, cuDNN, C++ Build Tools, FFMPEG & Git for AI Applications
Furkan Gözükara
Furkan Gözükara

Posted on

How to Set Up Python, CUDA, cuDNN, C++ Build Tools, FFMPEG & Git for AI Applications

Tutorial Link : https://youtu.be/DrhUHnYfwC0

This tutorial is essential for anyone looking to utilize AI, generative AI applications, and open source programs on their computer. I'll guide you through the proper installation of specific Python versions, demonstrate how to alternate between different Python versions, and explain the installation and switching process for various CUDA and cuDNN versions based on individual application requirements. Additionally, I'll cover the installation of Microsoft Visual Studio and C++ Build Tools (Compile) at the correct version for applications that need compilation on your system, such as InsightFace or XPose.

Video Resources

🔗 Tutorial Instructions Page (Publicly Available - No Paywall) ⤵️
▶️ https://www.patreon.com/posts/111553210

🔗 SECourses Discord Server for Comprehensive Support ⤵️
▶️ https://discord.com/servers/software-engineering-courses-secourses-772774097734074388

🔗 SECourses Reddit Community ⤵️
▶️ https://www.reddit.com/r/SECourses/

🔗 SECourses GitHub Repository ⤵️
▶️ https://github.com/FurkanGozukara/Stable-Diffusion

🔗 Previous Essential Tutorial (Highly Recommended) ⤵️
▶️https://youtu.be/-NjNy7afOQ0

Recommended Versions as of September 30, 2024

Python 3.10.11
CUDA 12.4
cuDNN 8.9 - most recent version - manually incorporated into system path
C++ Build Tools LTSC 17.8 - latest version
After viewing this tutorial, you should be able to use various AI applications without issues, including Automatic1111 SD Web UI, Kohya, SUPIR, IDM VTON, LivePortrait, FaceFusion 3, Rope Pearl, Rope Next, Roop, InstantID, Forge Web UI, Fooocus, CogVideoX, SwarmUI, ComfyUI, Stable Diffusion, SDXL, SD3, ControlNet, and more. This covers most popular open-source or proprietary AI applications that can be run locally on Windows.

This video also addresses how to resolve errors like:

onnxruntime\python\onnxruntime_pybind_state.cc:866 onnxruntime::python::CreateExecutionProvider Instance CUDA_PATH is set but CUDA wasnt able to be loaded. Please install the correct version of CUDA and cuDNN as mentioned in the GPU requirements page

Video Segments

0:00 Introduction to the tutorial
2:00 Verifying current Python, FFMPEG and CUDA versions
2:28 Correctly installing Python 3.10 and 3.11 in appropriate folders with proper configuration
4:25 Installing py launcher and its importance
4:56 Changing default Python version to 3.10 after installing both 3.10 and 3.11
5:10 Using Edit system environment variables to set default applications correctly
6:15 Properly installing Microsoft Visual Studio for on-device library compilation
7:33 Uninstalling previously installed applications before proper reinstallation
8:18 Correctly installing CUDA 11.8 and 12.4 on Windows for AI applications
12:53 Switching between default CUDA 11.8 and CUDA 12.4
13:33 Downloading and installing the latest FFMPEG and setting its system path
16:03 Installing the latest cuDNN libraries
17:23 Accurately installing the correct C++ Compile Build Tools compatible with CUDA 11.x and 12.x
19:12 Checking and removing previously installed C++ Compile Build tools
20:30 Downloading and installing Git for repository cloning
21:31 Installing and running LivePortrait AI on Windows locally, including animal Live Portrait
23:28 Resolving frozen CMD issues, paused application installs or executions
24:04 Exporting CMD installation logs on Windows 11 for error reporting
24:49 Fixing the "CUDA_PATH is set but CUDA wasn't able to be loaded" error
25:09 Installing older cuDNN 8.9.7 to address the CUDA_PATH error
27:34 Installing and using nvitop to monitor GPU and VRAM usage
28:56 Running the LivePortrait animal version for animating animal faces
29:55 Installing and running FaceFusion 3 locally on your computer
30:31 Identifying and resolving Git-related cloning errors
33:31 Solving DLL-related ONNX Runtime errors while running FaceFusion 3
33:50 Addressing ONNX Runtime errors by copying cuDNN 9.4 DLL files
34:38 Testing Face Fusion 3 with provided demo materials

Top comments (0)