DX-M1M
Ultra-Сompact M.2 AI Module
· 25 TOPS (INT8) AI Performance
· PCIe Gen3 x2 (Supports Gen 1/2/3 & x1/x2)
· Integrated 2GB LPDDR4x (4266 MT/s)
· 3W (Typical) Power Consumption
· M.2 2242 (M+B Key), 22 x 42mm
· Windows 11/10, Debian-based Linux, Yocto Project and Docker
· x86 and ARM-based Architectures
DX-M1M AI Module
DEEPX DX-M1M AI Module
Unlock the full potential of edge AI with a high-performance acceleration module powered by the DEEPX DX-M1M.
25 TOPS AI Computing Power
Powered by the DX-M1M chip, DeepX-M1M delivers 25 TOPS (INT8) of low-power local AI inference.
DX-M1M Acceleration Chip
Powered by the DEEPX DX-M1M AI acceleration chip, custom-built for localized edge computing scenarios.
Efficient INT8 Inference
Supports highly efficient INT8 neural network inference, achieving an optimal balance between performance and power consumption.
M.2 2242 Module Design
Utilizes a standard M.2 2242 form factor. The compact layout allows for rapid integration and streamlined product deployment.
Low-Power Edge Deployment
Engineered for low-power operation on edge devices, making it ideal for long-term, stable deployment.
ARM / x86 Platform Compatibility
Adaptable to mainstream computing platforms including ARM and x86, significantly enhancing system integration flexibility.
Visual Recognition & Object Detection
Perfectly suited for vision-based AI applications such as image recognition, object detection, and video analytics.
Industrial Vision & Robotics
Tailored for industrial vision, robotics, and smart gateways, driving the intelligent upgrade of edge devices.
High-Performance Acceleration AI Module
Designed for seamless integration into embedded and industrial systems, this compact AI module delivers up to 25 TOPS of AI performance while maintaining exceptional energy efficiency at just ~3W typical power consumption.
Application Scenarios
Suitable for a wide range of AI and edge computing applications, including robotics, private deployment of large language models (LLMs), edge AI computing, smart city infrastructure, smart home systems, and intelligent drone platforms.
Robots
Private Deployment of Large Models
Edge Computing
Smart City
Smart Home
Drone
Product Display
Specifications
| DX-M1M AI Module | ||
|---|---|---|
| Basic Specifications | AI Performance | 25 TOPS (INT8) |
| Host Interface | PCIe Gen3 x2 (Supports Gen 1/2/3 & x1/x2) | |
| Memory | Integrated 2GB LPDDR4x (4266 MT/s) | |
| Power Consumption | 3W (Typical) | |
| Operating Temperature | -40 ~ 85°C (Industrial) | |
| Thermal Solution | Heatsink (Option) | |
| Form Factor | M.2 2242 (M+B Key), 22 x 42mm | |
| OS Support | Windows 11/10 Debian-based Linux (Ubuntu 24.04/22.04/20.04 LTS) Yocto Project and Docker | |
| AI Frameworks | Ultralytics, PyTorch, TensorFlow, ONNX, Keras | |
| System Support | x86 and ARM-based Architectures | |
Customization
The Kiwi Pi team, bringing over 18 years of expertise in product design, research and development, and manufacturing, offers services including hardware, software, complete machine customization, and ODM/OEM.