Coral Edge TPU ML Accelerator Modules
Coral Edge TPU ML (Machine Learning) Accelerators are PCIe devices that allow easy integration of the Edge TPU coprocessor into existing systems to enable scaling to production hardware systems. Form factors are offered in half-size mini and M.2 (A+E, B+M, E). Supported hosts include OS: Debian Linux and/or Windows 10. All Coral Edge TPU Accelerators support TensorFlow Lite machine learning frameworks.
Embedded in all of the accelerators is the Coral Edge TPU coprocessor. The TPU is a small ASIC tensor processing unit designed by Google that provides high-performance ML inferencing for low-power devices. It can execute state-of-the-art mobile vision models such as MobileNet V2 at almost 400FPS in a power-efficient manner. An individual Edge TPU can perform 4 trillion (fixed-point) operations per second (4 TOPS), using only 2 watts of power (2 TOPS per watt).
Applications
- Embedded platforms
- Mini-PCs
- Industrial gateways
- IoT devices
- Workstations
Resource
Accelerator Product Selector
Coral TPU M.2 Edge Accelerator A+E Module
Small ASIC provides ML inferencing with low power using the Edge TPU coprocessor, keyed for A+E.
Coral TPU M.2 Edge Accelerator B+M Module
Small ASIC provides ML inferencing with low power using the Edge TPU coprocessor, keyed for B+M.
Coral TPU M.2 Dual Edge Accelerator Module
Includes two Edge TPU ML accelerators, each with a PCIe Gen2 x1 interface.
Coral TPU Mini PCIe Edge Accelerator Module
Brings the Edge TPU coprocessor to existing systems and products.
Coral TPU Edge SMT Accelerator Module
Solderable multi-chip module enables easy integration into existing systems.
Learn More
Murata Connectivity Modules
Simplifies wireless development by lowering the RF expertise needed to enable various applications.