AI System on Module
High-Performance AI Compute Module for Next-Generation Edge Intelligence
The AI System on Module ( AI SoM) is a high-performance AI compute module designed for next-generation edge intelligence devices. Built on an advanced multi-core processor architecture with an integrated NPU, the module delivers flagship-level AI inference, image processing, and system flexibility under compact size and, so it can serve as an ideal platform for intelligent vision systems and edge IoT devices.
Key Value Propositions
AI Acceleration Built-In
4K/8K Multimedia Processing
Industrial-Grade Reliability
Modular Design for Faster Development
Flagship-Level AI Acceleration
- Built-in high-efficiency NPU, supporting INT8 / INT16 / FP16 precision for a wide range of AI workloads
- Optimized for high-throughput inference with low power consumption, ideal for real-time edge computing
- Seamless compatibility with major AI frameworks (TensorFlow, ONNX, PyTorch via RKNN) for easy model deployment
- Optimized for vision AI tasks, delivering strong performance in detection, classification and segmentation
Advanced Multimedia & Display
- Up to 8K decoding and 4K encoding, delivering premium-quality video processing
- Supports multi-display output, ideal for video walls, dashboards and retail signage
- Multiple display interfaces available (HDMI, eDP, MIPI-DSI, DP) for various screen types
- Advanced ISP pipeline enhances camera image quality, ideal for machine vision scenarios
Rich High-Speed Interfaces
- Multi-channel MIPI-CSI camera interfaces for multiple high-resolution cameras
- High-speed expansion via PCIe, USB 3.x and SATA for NVMe SSDs, accelerators and peripherals
- Gigabit Ethernet and network module support for reliable edge connectivity
- Highly scalable I/O architecture, enabling custom system configurations
Industrial-Grade Reliability
- Industrial-grade wide-temperature support (e.g., -40°C to +85°C) for tough deployment environments
- Robust board-to-board connector design ensures reliability under vibration and shock
- Supports continuous 24/7 operation, ideal for mission-critical applications
- Built-in protection mechanisms including ESD protection, thermal management and power stability
Flexible OS & Ecosystem Support
- Supports Linux / Android platforms, covering a wide range of application needs
- Complete SDK, drivers and sample codes for rapid application development
- Compatible with mainstream AI tools including RKNN, OpenCV, TensorFlow Lite and more
- Comprehensive software ecosystem including BSP, inference engine, image libraries and debug tools
Modular Design & Easy Integration
- Standardized SoM + customizable carrier board architecture for product scalability
- Reference carrier designs reduce risks in high-speed signal development
- Developer kits, debug tools and complete documentation accelerate prototyping and mass production
- High reusability enables a unified compute platform across multiple product lines
Which AI Module is Right for You?
 |  | |
| نموذج | EC3588-C | EC3576-C |
| وحدة المعالجة المركزية | 4× [email protected] + 4× [email protected] | 4× [email protected] + 4× [email protected] |
| وحدة معالجة الرسومات | ARM Mali-G610 MC4 | ARM G52 MC3 |
| AI Engine | 6TOPS NPU | 6TOPS NPU |
| كبش | 16GB LPDDR4 | 8GB LPDDR4 |
| ذاكرة للقراءة فقط | 64 جيجابايت eMMC | 256GB UFS |
| Video Codec | • 8K@60fps H.265/H.264/AV1/AVS2 Decoder • 8K@30fps H.264/H.265 Encoder | • 8K@30fps H.264/H.265/VP9/AV2/AVS2 Decoder • 4K@60fps H.264/H.265 Encoder |
| عرض | • Built-in eDP/DP/ HDMI2.1/MIPI display interface，support multiple display engine max to 8K@60fps • Supports multi-screen display with 8K@60FPS max | •DisplayPort/MIPI/eDP/HDMI/RGB/EBC , multiple displays with different sources. |
| Camera | • Dual 16M Pixel ISP with HDR&3DNR • Multiple MIPI CSI-2 and DVP interface | • 16M ISP with HDR (up to 120dB) • MIPI CSI-2 (CDPHY=1*4-lane, DPHY=2*4-lane/4*2-lane), DVP |
| High-Speed Interface | PCIe3.0/PCIe2.0/SATA3.0/RGMII/TYPE-C/USB3.1/USB2.0 | PCIe2.1/SATA3.0/RGMII/USB3.0 |
| الموارد → | الموارد → | |
| ورقة البيانات → | ورقة البيانات → |