Google Coral GPIO System on Module (SOM) 2GB RAM 8GB eMMC
The Google Coral 2GB GPIO SOM can be used in building different systems on a small form factor with fast ML inferencing. NXP iMX 8M SoC is present in this GPIO SOM. This SoC is also present on the Coral Dev Board. Google Edge TPU coprocessor, Wi-Fi, Bluetooth, eMMC, and LPDDR4 RAM are present in this SoC.
What is Google Edge TPU?
Google Edge TPU has been designed by Google to speed up the TensorFlow Lite models. Power is used efficiently with this small ASIC. This is a 4TOPS ASIC and it just requires 2 Watts of power. This shows that per 2TO 1 Watt is used. You can get benefited from Google Edge TPU in many apps like MobileNet v2 that executes mobile vision models. Data privacy is regulated by the Google Edge TPU and inactivity is reduced. The requirement for a continuous internet connection is reduced by this on-device ML processing technique.
Specifications of Google Coral SOM
- SoC: NXP iMX 8M | ARM Cortex-A53 Quad-Core with Cortex-M4F
- Vivante GC7000 Lite 2D/3D GPU and VPU
- Coprocessor: Cryptographic Coprocessor
- Google Edge TPU Machine Learning Accelerator
- Bluetooth: 4.2
- Wi-Fi: 2.4/5 GHz 802.11b/g/n/ac 2x2 MIMO
- Ethernet: Gigabit Ethernet
- 8GB eMMC flash
- M2GB LPDDR4 RAM
- USB Interface: USB 3.0
- MIPI CSI 2
- MIPI DSI
- X95 (max.) GPIO including SAI, PWM, UART, I2C, SPI, and SDIO
- Dimensions: 48mm x 40mm
Benefits of Coral System on Module
- Complete Linux System (also runs Mendel)
- Low-Power and high-speed Machine Learning (ML) Inferencing (4 TOPS at just 2 Watts)
- Small 48mm x 40mm footprint