The Radxa AICore DX-M1M is a compact, energy-efficient M.2 edge AI acceleration module centered on the DeepX DX-M1M neural processing unit (NPU), delivering up to 25 TOPS (INT8) while consuming just 3W.
This module is crafted for industrial robotic arms, autonomous mobile robots (ARM), edge servers, drones, and AIoT devices, offering top-tier AI and ML capabilities without excessive power usage. It connects via a PCIe Gen3 x2 interface and is compatible with both x86 and Arm systems, such as the Raspberry Pi 5 and Radxa ROCK SBCs.
AICore DX-M1M Key Features:
- AI Accelerator – DeepX DX-M1M NPU offering up to 25 TOPS
- System Memory – 1GB LPDDR4X @ 4266 MT/s (on-chip, up to 8GB supported by DeepX)
- Storage – 1Gbit QSPI NAND / NOR flash
- Host Interface – PCIe Gen 3.0 x4 (supports Gen 1/2/3 and x1/x2) via M.2 M + B Key
- Power Consumption – 3W (Typical)
- Dimensions – 42 x 22 mm (M.2 2242 form factor); supports M.2 2280 slots with an adapter
- Temperature Range
- -25°C to 65°C (Stable; non-throttling)
- 65°C to 85°C (With thermal protection; throttling)
