AI
CORAL DEV BOARD
• High-speed and low-power ML
inferencing (4 TOPS @2W)
• A complete Linux
system (running
Mendel, a Debian
derivative)
• Prototyping and
evaluation board for
the small Coral SoM
(40 x 48 mm)
LEARN MOREu
GOOGLE CORAL DEV BOARD
DEMO BUILT WITH I.MX 8M
APPLICATIONS PROCESSOR
performing 4 trillion operations (tera-operations) per second
(TOPS), using 0.5 watts for each TOPS (2 TOPS per watt).
The Coral SOM, based on NXP's power efficient i.MX 8M
applications processor, pairs the Edge TPU with 1GB LPDDR4
RAM, 8GB eMMC flash memory, a secure element, Wi-Fi, and
Bluetooth 4.2 in a small module. The Coral
Dev Board TPU's small form factor enables
rapid prototyping covering internet-of-things
and general embedded systems that demand
fast on-device ML inference. Users can then
quickly move from prototyping to production
using the SOM combined with their custom
baseboard using board-to-board connectors.
For existing designs and scenarios where you
might want to use multiple Edge TPUs we offer
the Coral Accelerator with both USB and PCIe
interfaces. These are perfect for adding ML
inferencing acceleration to existing embedded
systems as well as to desktops and servers.
Coral helps you give devices local AI
processing capabilities that increase privacy,
reduce latency, and are connectivity agnostic in
a powerful, yet power-efficient package.
■
27
