Fanless GPU Edge System for Multi-Camera AI

Industrial AI applications such as image processing, quality control and autonomous systems require high computing power directly at the point of data generation. The GT-92GC from Acceed GmbH addresses this requirement with a fully fanless 19-inch rack system that integrates CPU and GPU processing, multi-camera connectivity and industrial-grade thermal design for edge computing environments.

Edge Computing Under Industrial Conditions
In manufacturing, robotics and autonomous platforms, real-time data processing reduces latency and limits dependence on cloud infrastructures. Processing image and sensor data locally lowers bandwidth demand and can reduce system costs within a digital supply chain architecture.

Edge systems deployed in these environments must combine high performance with resistance to dust, vibration and wide temperature ranges. The GT-92GC is designed for continuous industrial operation, supporting operating temperatures from −25 °C to +55 °C when configured with a 35 W processor.

The system is based on Intel’s 12th to 14th generation platform using the LGA1700 socket and supports CPUs with 35 W or 65 W thermal design power (TDP). GPU acceleration is provided by an NVIDIA RTX 2000 Ada with a 70 W power envelope, enabling AI inference and parallel image processing without active cooling.

Passive Thermal Architecture
Unlike conventional GPU-equipped industrial PCs, the GT-92GC uses a fully passive cooling system. Heat pipes and an extended fin structure dissipate thermal loads from both CPU and GPU without fans.

In harsh industrial environments, cooling fans are a frequent failure point. Eliminating active cooling components reduces mechanical wear and lowers maintenance requirements. The system’s thermal design is dimensioned to maintain CPU and GPU operation within specification under simultaneous maximum load, contributing to longer service life and lower total cost of ownership (TCO).

High-Density Camera Integration
Multi-camera systems are widely used in automated inspection, robotic guidance and autonomous vehicles. The GT-92GC integrates eight Gigabit Ethernet ports with PoE+ support according to IEEE 802.3at.

At an input voltage of 24 V, the PoE+ subsystem provides up to 100 W total power for connected devices, allowing cameras to be powered directly via Ethernet cabling. An additional Gigabit Ethernet port, four USB 3.2 Gen2 interfaces with 10 Gbit/s data rates, and optional serial, CAN bus and digital I/O interfaces expand integration options.

The architecture enables simultaneous connection of up to twelve cameras eight via Ethernet and four via USB supporting high-throughput inspection or sensor fusion tasks typical in AI-driven industrial automation.

Storage and Expansion Capabilities
For local data processing and buffering, the system includes an M.2 interface for NVMe SSDs with transfer rates of approximately 7 GB/s. Two 2.5-inch SATA drive bays allow additional storage capacity, including RAID configurations for redundancy or local data backup.

Two mPCIe slots with internal SIM card holders support modular expansion, such as mobile communication, Wi-Fi or additional I/O modules. Memory capacity scales up to 64 GB DDR5 RAM, aligning with current requirements for AI inference and high-resolution image analysis.

Application Context
The GT-92GC is positioned for applications requiring GPU-accelerated AI inference at the edge, including automated optical inspection, robotics, autonomous systems and sensor data fusion. By combining passive cooling with GPU performance in a rack-mounted format, the platform addresses a segment where thermal reliability and processing density must coexist.

Acceed GmbH, headquartered in Düsseldorf, Germany, distributes the GT-92GC in customer-specific configurations, including tailored power supplies and connection accessories for volume deployments. The company operates under an ISO-certified quality management system and supplies industrial computing, communication and signal processing solutions for automation, test management, quality assurance, and research and development environments.

www.acceed.com