Google launching Coral Accelerator Module at CES2020

Google is going to launch a new set of Coral Accelerator Hardware for the AI at the edge applications. Both are using Edge TPU Co-processor technology from Google.

Coral Accelerator module is chip scale miniature 15×10 mm SMT module with PCIe/USB 2.0 interface. On the other hand, the Dev Board mini is a full fledged Single board computer having MediaTek Quad core Processor with 8GB eMMC, 2GB RAM and other peripherals.

About Coral

Coral technology provides a complete platform for accelerating neural networks on embedded devices. At the heart of our accelerators is the Edge TPU co-processor. It’s a small-yet-mighty, low-power ASIC that provides high performance neutral net inference.

Google launching Coral Accelerator Module at CES2020 1
Google Coral Accelerator Module

The onboard TPU Co-processor is capable of 4 trillion operations per second (TOPS), using 0.5 watts for each TOPS (2 TOPS per watt). For example, it can execute state-of-the-art mobile vision models such as MobileNet v2 at 400 FPS, in a very power efficient manner.

Coral Accelerator Module Spec

Dimensions15.0 x 10.0 x 1.5 mm
ChipsetGoogle Edge TPU and PMIC
Mounting typeSMT, 120-pin LGA
Serial interfacePCIe Gen 2; or USB 2.0
Control interfaceI2C (optional)
Google launching Coral Accelerator Module at CES2020 2
Google Coral Dev Board Mini

Dev Board Mini Features

CPUMediaTek 8167s SoC (Quad-core Arm Cortex-A35)
GPUIMG PowerVR GE8300 (integrated in SoC)
ML acceleratorGoogle Edge TPU coprocessor
Flash memory8 GB eMMC
WirelessWi-Fi 5 (802.11a/b/g/n/ac); Bluetooth 5.0
Audio/video3.5mm audio jack; digital PDM microphone; 2.54mm 2-pin speaker terminal; micro HDMI (1.4); 24-pin FFC connector for MIPI-CSI2 camera (4-lane); 39-pin FFC connector for MIPI-DSI display (4-lane)
Input/output40-pin GPIO header; 2x USB Type-C (USB 2.0)

TensorFlow Lite

Developers need not to build models from the ground up. TensorFlow Lite models can be compiled to run on the Edge TPU.

AutoML Vision Edge

Easily build and deploy fast, high-accuracy custom image classification models to your device with AutoML Vision Edge.

News Source:

You can learn more about other AI Hardware available in the market in the article I have written some time ago.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.