The Forlinx FET1126Bx-S integrates a quad-core Arm Cortex-A53 CPU (up to 1.6 GHz commercial / 1.3 GHz industrial) paired with a 3 TOPS NPU supporting INT8/INT16 mixed-precision operations, enabling efficient execution of complex AI models such as personnel recognition, object detection, behavior monitoring, and fire or smoke alarms directly at the edge without cloud dependency.
The need for compact and cost-effective AI computing modules is growing in the fields of smart manufacturing, public safety, logistics, and infrastructure as AI deployment continues to transition from centralized cloud processing to distributed edge-side intelligence.
In order to improve real-time responsiveness and lessen reliance on bandwidth, many organizations are increasingly pursuing local AI inference. In this context, scalable and low-power edge modules are emerging as crucial components for industrial automation and AI-driven devices.
Forlinx Embedded Technology to announce the official availability of the FET1126Bx-S (Rockchip RV1126B/RV1126BJ) Industrial System on Module (SoM) in response to these changing market demands.
This compact module, designed specifically for edge-side AI computing, provides a high-performance, affordable solution for intelligent recognition and real-time video analysis in complex environments.
Edge-Optimized Architecture for Real-Time AI
The Forlinx FET1126Bx-S integrates a quad-core Arm Cortex-A53 CPU (up to 1.6 GHz commercial / 1.3 GHz industrial) paired with a 3 TOPS NPU supporting INT8/INT16 mixed-precision operations, enabling efficient execution of complex AI models such as personnel recognition, object detection, behavior monitoring, and fire or smoke alarms—directly at the edge without cloud dependency.
For rapid deployment, the RKNN toolchain ensures compatibility with major frameworks such as TensorFlow, PyTorch, Caffe, and MXNet.
Compared to the previous generation, the new FET1126Bx-S delivers significant upgrades in CPU/NPU processing power and transitions to Linux 6.1, ensuring a more powerful and seamless AI development journey.
High-Performance Multimedia Pipeline
-
Video & ISP
H.265/H.264 4K hardware decoding; parallel encoding up to 12 M @ 30 fps
12 Mbps ISP and 8 Mbps AI-ISP supporting HDR, 3DNR, fisheye correction, and dehazing
-
Display Interfaces (single display output, up to 1920×1080 @ 60 fps)
4-lane MIPI-DSI, 1.5 Gbps per lane for FHD panels
BT.656 / BT.1120 digital output for SD/HD/FHD displays with embedded sync
24-bit parallel RGB, up to 1920×1080 @ 60 fps
-
Camera Interfaces
Up to 2 × MIPI-CSI, 4-lane MIPI D-PHY v1.2 up to 2.5 Gbps/lane (MIPI-CSI/LVDS/Sub-LVDS)
DVP parallel camera input up to 150 MHz, compatible with BT.601 / BT.656 / BT.1120
Rich Connectivity in an Ultra-Compact Form Factor
Only 40mm×40mm with 237 pins led out via stamp hole and LGA connection. (140 stamp-hole pins at 1.0 mm pitch + 97 LGA pins at 1.27 mm pitch)
All functional pins of the RV1126B/RV1126BJ are fully led out, including display interfaces (MIPI-DSI, RGB LCD), network interfaces (Gigabit RGMII, 100Mbps Ethernet), and various peripherals (CAN FD, UART, SPI, I2C, PWM, ADC, etc.). The GPIO layout is fully compatible with the Raspberry Pi 40-pin standard.
Industrial-Grade Design and Reliability
Designed for long-term deployment, the FET1126Bx-S supports continuous 24/7 operation across ?40°C to +85°C, with LPDDR4 memory providing lower power consumption and improved resilience for different industrial environments.
Configurable LPDDR4 memory(1GB, 2GB, or 4GB) and eMMC storage options tailored to application needs: Fixed 64GB eMMC on the commercial variant and 8GB/16GB/32GB eMMC options on the industrial variant.
Each module undergoes rigorous stability and reliability testing in the Forlinx Embedded Laboratory, supporting 10–15 years of product longevity to ensure consistent long-term supply for industrial applications.
Target Applications
Designed for the next generation of smart infrastructure, the FET1126Bx-S is the ideal choice for:
- Smart Industries: Real-time production monitoring and equipment inspection.
- Smart Parks & Construction: Personnel safety (helmet detection), area intrusion, and smoke/fire alarms.
- Edge Computing & AIoT: Embedded vision and local AI inference for diverse IoT sensors.
Accelerate Your Time-to-Market
To help developers jumpstart their projects, Forlinx offers the OK1126Bx-S Evaluation Board. This production-ready platform features dual network ports, 2x MIPI-CSI, and a Raspberry Pi compatible 40-pin GPIO header, allowing you to bypass complex hardware design and focus entirely on your software application.
Availability
To support early-stage evaluation and design verification, Forlinx Embedded is currently offering a limited allocation of free engineering samples to qualified industrial customers and solution partners.
