Asus Launches Tinker Edge T SBC with Google Edge TPU
The board features an Edge TPU with NXP’s i.MX 8M quad-core Cortex-A53 processor, and Cortex-M4 real-time core and GC7000 Lite 3D GPU.
Asus first displayed the Tinker Edge T and CR1S-CM-A single-board computers back in June of last year during the Computex conference, which integrates Google’s Edge TPU for AIoT applications. While the status of the CR1s-CM-A industrial SBC is currently unknown, Asus has recently launched the Tinker Edge T, which features the same Coral SoM found on Google’s Coral development board.
Asus describes the Tinker Edge T as “specially designed for AI applications. It features the Google Edge TPU, a machine learning (ML) accelerator that speeds up processing efficiency, lowers power demands, and makes it easier to build connected devices and intelligent applications. With this onboard ML accelerator, Tinker Edge T is capable of performing four tera-operations per second (TOPS) using only 0.5 watts per unit of computation.” The board is optimized for TensorFlow Lite as well, making it easy to compile and execute common machine learning models for any number of applications.
On the hardware end, the Tinker Edge T packs NXP’s i.MX 8M SoC with quad-core Arm Cortex-A53 processor with a Cortex-M4F real-time core, GC7000 Lite 3D GPU, and an Edge TPU coprocessor. It also features 1Gb of LPDDR4 RAM and 8Gb of eMMC Flash, plus Wi-Fi 2×2 MIMO (802.11b/g/n/ac 2.4/5GHz) Bluetooth 4.2 for wireless connectivity options. Rounding out the board are a micro SD card slot, Gigabit Ethernet port, MIPI DSI connector (HDMI output up to 4K via CEC support), a pair of MIPI CSI interfaces for stereoscopic camera applications, two USB 3.2 Gen 1 ports, USB-C 3.2 Gen1 port, and 40-pin color-coded header.
The Tinker Edge T runs Mendel Linux (a lite version of Debian), which can be found running on Google’s Coral SBCs, and is an ideal platform for computer vision applications (denoted by the MIPI CSI interfaces). The Tinker Edge T is now available on Protovantage for $168 (although it shows a dual-processor motherboard), Physical Computing Lab for about $200, and Connection for $212, although the show the board as currently out of stock.