Resources
Radxa AICore DX-M1 - High-Efficiency Edge AI Acceleration Module For Computer Vision And Embedded Applications
The Radxa AICore DX-M1 is a high-performance M.2 AI acceleration module designed for computer vision, edge computing, and machine learning inference. It delivers up to 25 TOPS AI computing power while maintaining low 3 to 5W power consumption.
The Radxa AICore DX-M1 uses DEEPX DX-M1 technology with IQ8-based quantisation to provide high-precision AI processing. It is well-suited for smart cameras, security surveillance, autonomous mobility, robotics, and embedded-AI applications.
With a standard M.2 2280 M-key form-factor and PCIe Gen 3 x4 interface, the Radxa AICore DX-M1 integrates easily with compatible single-board computers such as Radxa ROCK 5B and ROCK 5B Plus.
This module is ideal for applications such as smart cameras, surveillance systems, autonomous mobility, and edge-based analytics.
Key Features:
- High AI Computing Power: Up to 25 TOPS for real-time computer-vision and AI inference tasks
- Energy-Efficient Design: Operates at 3 to 5W for optimised power usage in embedded systems
- High-Precision Processing: IQ8-based quantisation ensures FP32-level accuracy
- Standard Interface: M.2 2280 M-key form-factor for easy integration
- High-Speed Connectivity: PCIe Gen 3 x4 interface for fast data transfer
- Framework Compatibility: Supports PyTorch, ONNX, and TensorFlow
- OS Support: Compatible with Ubuntu and Debian Linux distributions
- Onboard Memory: 4GB LPDDR5 with 1Gbit QSPI NAND Flash
- Debug Support: UART and JTAG interfaces for development and diagnostics
- Wide Application Use: Suitable for edge-AI, robotics, smart surveillance, and mobility systems
AICore DX-M1 Module
25 TOPs Super Strong AI Computing Power
