PC with Nvidia L4 GPU for industrial edge AI

Two slim doors provide access for the company’s Flex I/O expansion boards

Called MIC-770 V3, “this industrial edge solution is boasts 7,424 Cuda cores and 30.3Tflop/s with 24Gbyte GDDR6 GPU memory in a 72W power envelope for edge AI and graphics applications”, said Advantech. “This creates an excellent AIoT solution for diverse deep learning and edge inference applications in factory automation and autonomous vehicles.”
It is a modular PC that can be expanded through two front panel slots for extra IO, and with an external add-on called MIC-75M20 (added left), a 45 x 192 x 230mm  module with one PCIe x16 slot, a PCIe x4 slot, and two optional 40mm fans.

“Adding a PCIE-1674 vision frame grabber card enables the connection of four cameras when used in autonomous vehicles,” said Advantech. “Likewise, adding 4x GbE Flex I/O [part 98910770301] enables this controller to connect to lidar systems, enabling it to detect distances and shapes in its environment and then pass on data to the Nvidia L4 GPU for the rendering of 3D models based on graphics analysis.”

For remote management, the PC supports iBMC 1.2 hardware-based out-of-band (and in-band) access.

“In addition, enterprise IT and OT managers can remotely control the power, conduct remote system recovery via Acronis or hardware, trigger SSD recovery, and check operating statuses — even during software or OS failures,” according to the company. “These features address 90% of system failures without requiring users to dispatch maintenance crews to the site.”

Operation is in ambients across 0 – 40°C, where the GPU can be maintained below 70.5°C, it added.

Use is also foreseen in automated optical inspection, AI prediction in medical equipment, smart city video surveillance and transportation.