Microsoft and Arm partnered to address the costly challenge of learning new ML tools and to close the gap on end-to-end cloud-to-Emulation-to-Arm cores and NPU. Microsoft and Arm developed an Azure-hosted pipeline to select a ML model, to compile and optimize it with awareness of the IoT silicon target’s performance at the edge.

The session and demo shows the custom CI/CD pipeline for Cortex-A, Cortex-M, and Ethos NPU for developer and solution operators to leverage a virtual target to build a complete solution.