hollance/neural-engine
Community research documenting how to optimize ML inference performance on Apple's Neural Engine hardware for iPhones and iPads.

A documentation project exploring Apple Neural Engine (ANE) capabilities and limitations for running neural network models on iOS devices. The repository explains how to optimize Core ML models to leverage ANE acceleration over CPU and GPU, covering convolutions, matrix multiplies, and hardware-specific constraints. It provides guidance on achieving performance gains through trial-and-error experimentation rather than official Apple documentation.