Google and Murata have partnered to create the Coral Accelerator Module based on Google’s Edge TPU ASIC. The custom-designed module measures a tiny 10 mm x 15 mm x 1.5 mm while delivering up to 4 TOPS of performance at the expense of only 2 TOPS per watt of power consumption. This is thanks in part to a PMIC integrated into the Edge TPU chipset.
The 120-pin LGA surface-mount module offers advanced noise suppression and a simplified PCB design that can be soldered easily. Supported interfaces include PCIe Gen 2, USB 2.0, and I2C (optional).
The Coral Accelerator Module supports the TensorFlow Lite deep learning framework and AutoML Vision Edge computer vision modeling software.
The Coral Accelerator Module will be available for sale in early 2020.
For more information, visit www.coral.ai. To view performance benchmarks, visit coral.ai/docs/edgetpu/benchmarks.
About the Author
Follow on Twitter Follow on Linkedin Visit Website More Content by Brandon Lewis