Industrial edge solution with NVIDIA L4 GPU for industrial AI applications

Advantech has announced its GPU edge computing solution with NVIDIA L4 GPU NVQual validation – the MIC-770 V3 Modular IPC with MIC-75M20 expansion i-Module. This industrial edge solution is compatible with the NVIDIA L4 Tensor Core GPU, provides an impressive 7,424 CUDA cores and 24GB GDDR6 GPU memory, and consumes just 72W of power. When paired with the efficiency of the NVIDIA L4 GPU, the high-performance, highly scalable compact design of this IPC produces an outstanding AIoT solution for diverse deep learning and edge inference applications in factory automation, autonomous vehicles, AOI, and AI prediction in medical equipment, and smart city video surveillance and transportation.

Powered by the latest 12th Gen Intel Core i processor, the solution is designed as a high-performance open-source x86 platform that enables the rapid development of GPU-accelerated solutions. By adopting MIC-75M20, a two-slot expansion i-Module, this integrated solution is compatible with the NVIDIA L4 Tensor Core GPU with its small, energy-efficient, single-slot, low-profile 72W form factor. These attributes make it excellent for AI-based edge computing and inference. It produces energy-efficient universal acceleration for AI inference, video, and graphics applications; and is ideal for global deployment in enterprise, cloud, and edge applications with limited space.

The solution features a modular design that delivers enhanced functionality and flexibility for PCIe expansion via the company's i-Modules and Flex I/O in the front panel. Adding a PCIE-1674 vision frame grabber card enables the connection of 4 x cameras when employed in autonomous vehicles. Likewise, adding 4 x GbE Flex I/O (98910770301) allows this controller to connect to LiDAR systems, therefore, allowing 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.

It supports the iBMC 1.2 Edge Intelligent Solution to simplify the management of remote systems and edge devices. Indeed, by leveraging iBMC hardware-based out-of-band management technology, this WISE-DeviceOn solution can serve as an in-band and out-of-band management system, allowing the complete access, configuration, monitoring, and analysis and control of IoT network assets on a single centralised platform. Also, Enterprise IT and OT managers can remotely control the power (through reset/forced shutdown/power on/power off), conduct remote system recovery via Acronis or hardware, trigger SSD recovery, and check operating statuses – even during software or OS failures. These elements address 90% of system failures without needing users to dispatch maintenance crews to the site, therefore, drastically decreasing system downtime and operational costs.

When employed in harsh environments, the solution leverages an NVIDIA L4 GPU and supports wide operating temperatures (0-40C). It provides a dedicated server-grade air duct design that enhances the thermal management of the GPU. This cooling solution maintains a GPU IC operating temperature below 70.5C in environmental conditions up to 40C to provide superior performance and clock frequency residency. It leverages its advanced thermal design to provide 30.3 TFLOPs computing power with 72W of power consumption.