Asus debuts new mini PC and hobbyist boards for AI edge computing workloads

Asus debuts new mini PC and hobbyist boards for AI edge computing workloads

Asus hopes to tap into an edge computing market anticipated to be worth $9 billion by 2024. To this end, the Taiwan-based company unveiled a new Mini PC — the PN61T — as well as the Tinker Edge R and Edge T, which were first announced last November. Asus says they’re designed to work in tandem within enterprise environments to enable check-in via facial recognition and automated lighting, plus wireless presentations and persistent device monitoring.

All three new products will be available starting in 2020, according to the company.

The PN61T — which slots alongside the PN61V in Asus’ Mini PC lineup — features up to an Intel Core i7 processor and a built-in Google Edge tensor processing unit (TPU), which speeds up processing while lowering overall power consumption. The Edge TPU can execute multiple computer vision models at 30 frames per second (fps) or a single model (like MobileNet V2) at over 100fps, thanks in part to a lightweight version of Google’s TensorFlow framework, dubbed TensorFlow Lite. It’s also able to tap Google’s Cloud IoT Edge for data management and processing.

Elsewhere on the PN61T’s stackable chassis is a dual-microphone array and audio inputs, plus infrared sensors that allow the unit to be controlled wirelessly with a remote. Available ports include USB 3.1 Gen2 Type-C ports, a MicroSD card reader, a USB Type-C input for quick charging, and an HDMI input, in addition to a configurable port option with VGA, COM, DisplayPort, LAN, and Thunderbolt 3 connections.

As for the Tinker Board Edge series, it comprises the Tinker Edge T and Tinker Edge R — two single-board computers in a small factor that offer embedded designs with multi-core Arm Cortex processors. They both boast capable graphics chips and multimedia processors (that can handle up to 4K on the Tinker Edge R), and robust connectivity from Gigabit Ethernet to Wi-Fi and Bluetooth. Forty general-purpose input/output pins round out the I/O, complementing the NXP i.MX 8M system-on-chip (on the Tinker Edge T) and Rockchip RK3399 Pro (on the Trinker Edge R) and HDMI, MIPI-DSI, Dual MIPI-CSI, USB 3, and Type-C ports.

Each Tinker Edge computer comes with support for a fully open source kernel that plays nicely with multiple APIs, including OpenGL, Vulkan, OpenCL, OpenVX, Google’s AndroidNN, and Facebook’s Caffe and Caffe2.

“This latest lineup of … products [is] designed for real-life AI edge-computing applications, from AI-enabled consumer robotics and smart vending machines to AI-enabled video conferencing and camera setups,” wrote Asus in a press release. “Because the NPU is optimized for handling neutral networks, the Tinker Edge series is ideal for bringing artificial intelligence capabilities from the cloud to a local device while keeping user data local and secure.”

Source: Read Full Article