Industrial vision hardware should be sized to the workload. Many video analytics tasks can run directly on a modern CPU. As camera counts, model complexity or frame rates increase, so does the need for GPU acceleration.
The FPC-9309W-G5 can be configured accordingly, with Intel Core Ultra Series 2 processing and an NVIDIA RTX 50 GPU where required. Four independent 2.5 GbE ports and five PCIe slots become particularly valuable when the computer must handle several data streams and additional expansion cards.
Core Ultra Processing with Optional GPU Acceleration
The LGA1851 platform uses the Intel W880E chipset and accepts 35 W or 65 W Core Ultra Series 2 processors. Four DDR5 U-DIMM sockets support up to 192 GB of DDR5-5600 ECC or non-ECC memory, providing capacity for large image buffers, local datasets and several concurrent services.
At VISION 2026, ARBOR and SAIMOS demonstrate an intelligent video analytics workload running on the FPC-9309W-G5 entirely on the Intel processor.
For workloads with greater model complexity, more simultaneous streams or higher throughput requirements, the chassis supports one RTX 50 series GPU consuming up to 575 W and measuring up to 357.6 mm long. That power limit matches the Total Graphics Power of NVIDIA's RTX 5090 reference design. A dual-GPU build using cards rated at up to 300 W each is another supported option.
Add-in-board models can differ in length, thickness, power connector layout and cooling design. The exact card must therefore be checked against the complete computer configuration rather than selected by GPU designation alone.
Expansion and Network Resources for the Complete Application
Two PCIe x16 slots accommodate high-performance graphics cards. Three additional physical PCIe x8 slots, each electrically connected through four lanes, provide room for frame grabbers, network adapters, storage controllers or industrial I/O cards.
A wide GPU cooler may block an adjacent slot. Lane allocation, card clearance, power consumption and airflow must therefore be considered across the complete set of add-in cards.
Network connectivity is provided by four dedicated Intel WGI226LM controllers. The four 2.5 GbE interfaces can be assigned to separate camera, machine, plant and service networks. PoE is not included in the specified interface set, so connected cameras or other devices require separate power or an external PoE switch.
Local Storage and Front-Accessible I/O
Two M.2 M-key sockets accept NVMe SSDs, while two externally accessible bays accommodate 2.5-inch or 3.5-inch SATA drives. RAID 0, 1 and 5 are available depending on the drive configuration. Operating system files, AI models, recordings and temporary processing data can therefore be placed on separate storage devices.
The external interface set includes eight USB 3.2 Gen 2 ports with transfer rates up to 10 Gbit/s, four serial ports and 16-bit digital I/O. Two HDMI 2.1 and two DisplayPort 2.1 connectors provide a choice of display interfaces, with support for up to two independent displays at the same time.
Front-accessible connectors and drive bays simplify installation and service when the computer is mounted inside machinery or a control cabinet.
Power Budget and Thermal Limits
The FPC-9309W-G5 accepts a 24 to 48 V DC input. For sites supplied from AC power, ARBOR offers optional 600 W and 1,000 W AC/DC DIN-rail adapter kits. The power supply must be selected for the processor, GPU, storage devices and additional PCIe cards installed in the final build.
The operating temperature range also changes with the hardware configuration:
| Configuration | Manufacturer-specified operating range |
|---|---|
| 35 W TDP processor | −20 to +70 °C |
| 65 W TDP processor | −20 to +55 °C |
| System fitted with an NVIDIA RTX 5090 | −20 to +45 °C |
For operation with SSD storage, ARBOR specifies vibration resistance to MIL-STD-810G Method 514.6, Category 4, and shock resistance to Method 516.6 at 20 G. These figures apply to the manufacturer's defined test configuration. A fully populated computer requires a separate mechanical and thermal assessment that includes the GPU and every additional expansion card.
The metal wall-mount enclosure measures 220 × 380 × 320 mm and weighs 7.4 kg.
Specifying the System for a Project
The FPC-9309W-G5 is suitable for multi-camera machine vision, automated optical inspection, on-site video analytics, and demanding compute or control functions in automation and robotics. The final hardware should be based on the actual data flow and installation conditions rather than the maximum specification alone.
IPC2U will need the following information to prepare a suitable configuration:
- software stack and AI models
- number, resolution, frame rate and interfaces of cameras or other data sources
- intended GPU and additional PCIe cards
- storage capacity and RAID requirements
- available input power and maximum ambient temperature
Review the available options on the FPC-9309W-G5 product page or browse further Edge AI computing platforms. Project requirements can be submitted to the IPC2U team for configuration and quotation.
