Coral USB Accelerator
Works with Linux, Mac and Windows systems.
Speed up machine learning inferencing
The Coral USB Accelerator is a USB accessory that contains a specialized ASIC (Edge TPU) for acceleration of machine learning (ML) inferencing calculations.
The Edge TPU coprocessor is capable of 4 trillion operations per second, using only 2 Watts of power.
All major platforms - Debian Linux (including Raspberry Pi), macOS, and Windows 10 are supported.
Simply compile TensorFlow Lite models to run on the Edge TPU.
ML accelerator: Google Edge TPU coprocessor, supports high performance ML inferencing for TensorFlow Lite models
Connector: USB 3.0 Type-C (data/power)
Software support: Supports Linux, Mac and Windows on host CPU; Python 3.5 / 3.6 / 3.7
Dimensions: 65 mm x 30 mm
Model: WA1
NB: The USB connector is compatible with USB 2.0 but the inferencing speed will be much slower for USB 2.0.
Datasheet & Benchmarks
Performance TL;DR:
~ 20 - 40 x speed-up in inferencing time compared to Embedded CPU, depending on the model architecture, CPU & connection used. See link for more information.
Using an appropriate model, the reduced inferencing time with the Edge TPU will allow you to do real-time inferencing at 50 fps or more.
Google part numbers:
- G950-01456-01 (old part number)
- G950-06809-01 (all units purchased directly from Google starting October 2020)
Asus part number: 90AN0020-B0XAY0
Additional Legal Information
The products offered by Google are unrelated to the products offered under the CORAL trademarks owned by Orient Development Enterprises Ltd. in Taiwan.