Advantech introduces the MIC-770 V3 Industrial Edge Solution focused on industrial AI with support for NVIDIA L4 GPU

Advantech recently released a new GPU edge computing product, MIC-770V3 modular industrial computer, which can be paired with the i-Module series GPU expansion module MIC-75M20. This high-performance product is compatible with NVIDIA L4 core Gpus and supports 7,424 CUDA cores and 24GB GDDR6 GPU memory with a power consumption of only 72W. In addition, the product has been verified by NVIDIA L4 GPU NVQual, which is suitable for smart factory, automatic driving, automatic optical detection (AOI), AI prediction of medical equipment, smart city monitoring and traffic, etc., while supporting deep learning and edge reasoning, It also gives full play to its characteristics of high performance, flexible expansion and compact design

The MIC-75M20, a 2-slot i-Module expansion module compatible with NVIDIA L4 Core GPU, has the features of small size, power efficiency, single slot and 72W low power consumption. These features make it ideal for AI edge computing and reasoning, providing power-efficient, universal acceleration for AI inference computing, imaging and graphics applications that help enterprises deploy in the cloud, on the edge, and in space-constrained environments.

The MIC-770 V3 has a modular design with enhanced functionality and more flexibility through Advantech I-Modules and rich I/O interface design on the front panel. When used in autopilot, the PCIE 1674 Vision acquisition card can be connected to four cameras. Similarly, four GbE Flex I/O ports (98910770301) are connected to the Lidar system. After the distance and shape of objects in the environment are detected, the data is passed to NVIDIA L4 GPU, which flexibly realizes 3D model rendering based on graphics analysis.

The MIC-770 V3 supports the iBMC 1.2 Edge Intelligence solution, simplifying remote system and edge device management. By leveraging iBMC hardware's remote out-of-band (OOB) management technology, Advantech's WISE-DeviceOn solution can be used for both in-band and out-of-band management systems, enabling comprehensive access, configuration, monitoring, and analysis control of Internet of Things (iot) network assets in a single centralized platform. In addition, even if a software or operating system fails, enterprise IT and OT managers can still control the power remotely (by reset/forced shutdown/power on/power off, etc.), perform remote system recovery via Acronis or hardware, trigger SSD recovery, and check the running status. These features resolve 90% of system failures without having to send maintenance personnel to the facility, significantly reducing system downtime and operating costs.

The MIC-770 V3 uses NVIDIA L4 GPU and runs at temperatures ranging from 0 to 40℃ in harsh environments. When the temperature reaches 40°C, the dedicated server-class air duct design improves the heat dissipation processing of the GPU and keeps the GPU operating temperature below 70.5°C to ensure excellent performance and clock frequency residence. All in all, the MIC-770 V3 is designed for professional heat dissipation, delivering 30.3 TFLOPs while only consuming 72W.

Main characteristics

● NVIDIA certified NVIDIA L4 Tensor Core GPU

● Powerful performance with 7,424 CUDA GPU architecture, 30.3 TFLOPS, 24GB GDDR6 GPU memory, and power consumption of only 72W for edge AI and graphics programs

● Intel 12th Generation Core™i9/i7/i5/i3 Processor (LGA 1700)

● High-performance solutions support compact rugged design with support for Flex I/O, iDoor extension modules

● Advantech iBMC 1.2 Remote Management Solution (WISE-DeviceOn)

● Suitable for installation in space limited environment and harsh outdoor environment