DFRobot
DX-M1 AI Accelerator M.2 Module - 4GB LPDDR5, 25 TOPS
· MPN: DFR1252
Add dedicated edge AI acceleration to your host system with this compact DX-M1 M.2 module. It delivers 25 TOPS (INT8) inference performance in a standard M.2...
Add dedicated edge AI acceleration to your host system with this compact DX-M1 M.2 module. It delivers 25 TOPS (INT8) inference performance in a standard M.2 2280 form factor, with low 2W–5W power consumption for embedded and mobile systems.
The module uses an M.2 M-Key interface over PCIe Gen3 x4, with compatibility for x1 mode, making it suitable for x86 PCs, LattePanda boards, Raspberry Pi 5 setups via suitable HATs, and other x86/ARM platforms. Onboard 4GB LPDDR5 memory helps handle larger models, while 1Tbit QSPI NAND Flash is included for firmware storage.
Development is supported by the DXNN® SDK, which provides a workflow for model compilation, optimisation and runtime deployment. It supports common AI frameworks including PyTorch, ONNX, TensorFlow, TensorFlow Lite, Keras and XGBoost for applications such as robotics, visual SLAM, real-time video analytics, object detection and industrial inspection.
Included in the package is one DX-M1 AI Accelerator M.2 Module with 4GB LPDDR5.
Features:
- Edge inference: 25 TOPS (INT8) performance for complex neural networks and multi-stream video analysis.
- Low power operation: Designed to operate within a 2W-5W power envelope.
- M.2 integration: Standard M.2 M-Key (2280) format for broad platform compatibility.
- PCIe interface: Uses PCIe Gen3 x4 and is backward compatible with x1 mode.
- Development support: Backed by the DXNN® SDK for compilation, optimisation and runtime execution.
- Framework support: Supports PyTorch, TensorFlow, TensorFlow Lite, Keras and XGBoost, with ONNX workflow support.
- Industrial use: Suitable for robotics, visual SLAM, AI visual inspection, safety monitoring and autonomous systems.
Specifications:
- Processor Performance: 25 TOPS (INT8)
- Interface: M.2 M-Key, PCI Express Gen3 x4 (compatible with x1 mode)
- Memory: 4GB LPDDR5, 1Tbit QSPI NAND Flash
- Power Consumption: 2W ~ 5W
- Power Range: 3.3V±5%
- Framework Support: PyTorch, ONNX, TensorFlow, TensorFlow Lite, Keras, XGBoost
- Operating Systems: Windows 10/11, Ubuntu 20.04/22.04 LTS
- Operating Temperature: -25°C ~ 85°C (Throttling); 25°C ~ 65°C (Non_ Throttling)
- Product Dimensions: 22 mm x 80 mm x 4.1 mm/0.87 inch x 3.15 inch x 0.16 inch
A strong fit for makers and engineers adding local AI inference to robotics, embedded vision, industrial automation or edge computing builds.
Jargon buster
Plain-language definitions for the technical terms used above.
- edge computing
- Edge computing means processing data close to where it is collected, such as on the device itself, rather than sending everything to the cloud. This can reduce delays, internet dependence, and privacy concerns in sensor, camera, and robotics projects.
- M.2
- A compact edge-connector format commonly used to plug small modules into a carrier board without soldering. On this product it is the physical connector used by the MicroMod system, so compatibility with the matching processor board is important.
- TOPS
- TOPS means trillions of operations per second, often used to describe AI accelerator performance. It helps compare whether a computing module is suited to lightweight image recognition or more demanding neural-network workloads.
Supplier page — dfrobot.com
Supplier Description · 146.4 KB · Click any page to view full size
