Accelerator module performs high-speed inferencing for ML models

Coral Surface Mount Accelerator Module with Edge TPU, available now from Mouser, is a multi-chip module (MCM) developed to perform high-speed inferencing for machine learning (ML) models. It comprises the Edge TPU ML accelerator with integrated power control, and it may be connected over a PCIe Gen2 x1 or USB2 interface.

The Edge TPU is a small ASIC design that accelerates TensorFlow Lite models in a power-efficient manner. It can perform four trillion operations per second (4TOPS), using 2TOPS per watt. For example, one Edge TPU can execute state-of-the-art mobile vision models such as MobileNet v2 at nearly 400fps. This on-device ML processing lowers latency, improves data privacy, and stops the necessity for a constant internet connection.

Coral, Google's platform for on-device AI, provides hardware components, software tools, and pre-compiled models for building intelligent devices. Machine learning models includes embedded platforms, mini-PCs, and industrial gateways.