Profile a model on a neural accelerator with MLIA for VS Code
Introduction
Prepare your model and VS Code workspace
Prepare the Android workload
Capture NX profiling data
Analyze measured performance in VS Code
Next Steps
Profile a model on a neural accelerator with MLIA for VS Code
Who is this for?
This Learning Path is for machine learning or graphics developers investigating model performance on Arm neural accelerators (NX).
What will you learn?
After completing this Learning Path, you'll be able to perform the following tasks:
- Export a small neural upscaling model with Python debug information.
- Capture one NX dispatch with Android Scenario Runner.
- Inspect measured cycle costs in Model Explorer.
- Follow graph operations to Python source in Visual Studio Code (VS Code).
Prerequisites
Before starting, you'll need the following prerequisites:
- An x86-64 Ubuntu 24.04 LTS host with internet access and permission to install development tools
- Visual Studio Code 1.100 or later on the host
- An NX-capable Android device with USB debugging that meets the Arm ML Inference Advisor (MLIA) VS Code device requirements
- Python 3.12 with virtual environment support, Android Debug Bridge (
adb), curl, and tar on the host - Familiarity with Bash, adb, and PyTorch model export
Summary
This summary was drafted with an approved AI-assisted workflow and reviewed by Arm contributors before publication. Human technical review remains part of the process so the final page reflects engineering rigor, accuracy, and Arm editorial standards.
You’ll profile a neural upscaling model on an Android device with an Arm NX, then inspect measured cycle costs in MLIA for VS Code. First, you’ll export the VGF model and set up the extension. Next, you’ll prepare Scenario Runner and the device, then capture an NX dispatch. Finally, you’ll load the capture and matching VGF. You’ll inspect the cycle cost of the cascade, and trace an operation to Python source.