Apple ANE (Neural Engine)

Realizes: AI accelerator

Apple's Neural Engine inside A-series and M-series SoCs is a dedicated neural processing unit composed of hardware matrix multiply arrays and supporting SRAM/control pipelines that deterministically accelerate inference workloads on-device.

Examples

iPhone 12 and M1 Macs with Apple Neural Engine

iPhone 12 and Apple Silicon M1 Macs perform on-device inference workloads like image segmentation, speech recognition, and AR tracking by routing models through the ANE's matrix multiply fabric, keeping latencies predictable and local.

INFERENCE MATRIX_MULTIPLY low-latency mobile flagship low energy per operation