microsoft / microsoft/onnxruntime

[Build] ONNX Runtime Support for Cortex-M33 and Cortex-M7

Open
#23,822 2 comments 0 reactions 0 assignees View on GitHub
contributions welcome
Dominant language
C++
Stars
21.9k
Forks
4.2k
Avg merge
4d 11h
Merged PRs (30d)
184

Description

### Describe the issue

I successfully ran ONNX Runtime on Cortex-A55 for a basic matrix multiplication example by directly downloading the AArch64 ONNX Runtime library.

However, I want to try it on low-level microcontrollers, so I followed the [official ONNX Runtime build guide](https://onnxruntime.ai/docs/build/inferencing.html) to generate an ONNX Runtime library for ARM Cortex-M33 and Cortex-M7. I am unsure if running ONNX Runtime on Cortex-M series microcontrollers is feasible.

Since Cortex-M33 is part of the ARMv8-M family, does ONNX Runtime officially support it? If not, are there any recommended lightweight alternatives for running ONNX models on Cortex-M33/M7?

I would appreciate any insights or guidance on this

### Urgency

_No response_

### Target platform

M33 and M7

### Build script

I followed the [official ONNX Runtime build guide](https://onnxruntime.ai/docs/build/inferencing.html) to generate an ONNX Runtime library for ARM Cortex-M33

### Error / output

Couldn't able to build

### Visual Studio Version

_No response_

### GCC / Compiler Version

_No response_

Contributor guide

Open the contributing guide

Research direction

Read the official ONNX Runtime build guide and compare its documented targets with Cortex-M33 and Cortex-M7; reproduce the failed library build for those platforms. Done means establishing whether they are supported and documenting a recommended lightweight alternative if they are not.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
embedded-iot, machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.