Building BuddyGenAI on Windows for NVIDIA GPUs

This guide will walk you through the process of building BuddyGenAI on Windows for NVIDIA GPUs with CUDA acceleration.

This is the setup that I have been using so I would expect it to work for you as well. If you have any issues or need help with any details, please let me know so that I can update this guide.

Prerequisites

• Microsoft Visual C++ Redistributable x64
• Node v20
  • Install NVM for Windows
  • Run nvm install 20 and then nvm use 20, and restart your terminal
• CUDA Toolkit
  • Optional if you download and use the cudart* zip below.

Acquire Binaries

First, you'll need to download or build binary files that BuddyGenAI relies on. The latest versions should work, although you may need to check the date of BuddyGenAI's latest release to find the right version of any binaries that are not working.

You'll need the following project's binaries:

• llama.cpp
  • Download the latest release's llama-*-bin-win-cuda-cu12.2.0-x64.zip
  • Download cudart-llama-bin-win-cu12.2.0-x64.zip if you did not install CUDA Toolkit
• stable-diffusion.cpp
  • Download the latest release's sd-master-*-bin-win-cuda12-x64.zip
• Piper
  • Download the latest release's piper_windows_amd64.zip
• whisper.cpp
  • (not all releases have CUDA builds)
  • Download the latest release's whisper-cublas-12.2.0-bin-x64.zip

Now, create a binaries folder in the root of the project with a folder called cuda12 inside it. Extract the contents of the Piper zip into the binaries folder. Extract the contents of the rest of the zips into the cuda12 folder. You'll need to rename the *.exe files for llama.cpp, stable-diffusion.cpp, and whisper.cpp to match the expected names below.

The following is the expected structure of the binaries folder:

binaries/
├─ cuda12/
│  ├─ cublas64_12.dll
│  ├─ cublasLt64_12.dll
│  ├─ cudart64_12.dll
│  ├─ ggml.dll
│  ├─ ggml_shared.dll
│  ├─ llama.dll
│  ├─ llamacpp-main.exe
│  ├─ llamacpp-server.exe
│  ├─ stable-diffusion.dll
│  ├─ stable-diffusioncpp-sd.exe
│  ├─ whisper.dll
│  ├─ whispercpp-main.exe
├─ espeak-ng-data/
│  ├─ ...
├─ pkgconfig/
├─ espeak-ng.dll
├─ libtashkeel_model.ort
├─ onnxruntime.dll
├─ onnxruntime_providers_shared.dll
├─ piper.exe
├─ piper_phonemize.dll

Build BuddyGenAI

git clone https://github.com/parsehex/BuddyGenAI
cd BuddyGenAI
# IMPORTANT! Get the main branch:
git checkout main
npm install
npm run build:electron

After building completes, you can find a setup .exe in the electron-dist folder.