Welcome

Welcome to the AI docs for Qualcomm Dragonwing. Dragonwing is a family of IoT SoCs that offer high performance, advanced connectivity and efficient power - paired with fast GPUs and NPUs; and running a variety of operating systems (Ubuntu, Qualcomm Linux, Android and Windows). This makes them perfect for advanced edge AI workloads, or high-performance computer vision applications.

This documentation is where you’ll find all the information you need to set up your Dragonwing platforms, run your first AI or CV application, and how to run your existing workloads at full performance.

Note: This documentation is work in progress; and is written for development boards running Ubuntu 24.

Device setup

We support the following development boards. Follow the setup guide to prepare your development board and install Ubuntu 24:

• RB3 Gen 2 Vision Kit

• Rubik Pi 3

Then, setup Visual Studio Code (VS Code) to develop for your board remotely via SSH.

Running AI models

We support a wide range of ways to run AI models. Whether you use a platform like Edge Impulse, want to download pretrained models from AI Hub, or want to bring your own inference pipeline using TensorFlow Lite (LiteRT) or ONNX. Here's a full list of guides:

• Building your own AI models using Edge Impulse

• Find and run pretrained models on AI Hub

• Run LiteRT/TensorFlow Lite models on NPU

• Run ONNX models on NPU

• Run LLMs / VLMs on GPU using Llama.cpp

• Run LLMs / VLMs on NPU using Genie

• Transcribe audio using Whisper on NPU

• Running computer vision applications on GPU with IM-SDK