info@embecosm.com

A common perception is that all AI needs a great deal of compute power. While this is true when training models, and for inference of large language models, it is not true in many other cases. AI inference models for image detection and analysis, or for audio processing can be run successfully on the very smallest of processors. Such small, fast, low-power models are central to the success of edge AI.

Since September 2026, we have been publishing a series of posts, showing how the IREE MLIR tool flow can be used to bring up AI models on small bare metal microcontrollers. This post is a master post, bringing together all the individual posts for the convenience of the reader. It is updated, each time we create another post in the series.

The following posts have been published in this series:

• Part 1: Overview of IREE on a bare metal microcontroller
• Part 2: An open reference platform for IREE on a bare metal microcontroller
• Part 3: A matrix multiplication from MLIR down to RISC-V on one core

As each new post is published, we’ll extend this list.