LM5010 M.2

LM5010 M.2

Built on the standard M.2 2280 form factor, the LM5010 M.2 brings the powerful computing capabilities of the M50 into a gumstick-sized card. Its plug-and-play design enables compact devices to gain on-device AI capabilities with ease. Featuring low-power, fanless operation, broad operating-system compatibility and full-stack development support, it can be flexibly deployed in AI PCs, AI NAS systems, all-in-one conference devices and other equipment, delivering local large-model inference with enhanced data privacy and low-latency response.

Specifications

Dimensions

22×80mm

Interfaces

PCIe Gen4 ×4, M.2 M-Key

AI Accelerator

1× MOMAGIC ® 50

Physical Computing Power

160 TOPS @INT8
100 TFLOPS @bFP16

Typical power consumption

≤ 13W

Memory Capacity

12~48GB 192-bit LPDDR5/5X

Memory Bandwidth

153.6 GB/s

System Support

Win11 / Linux / Android

Software Support

Houmo Dadao™ Software Stack and Houmo Open-Source AI Ecosystem Software

Power Management

Multi-level power consumption modes, ultra-low power standby as low as 4mW

Operating Temperature

-20°C ~ +60°C

Cooling Solution

Custom heatsink, supporting active cooling / passive cooling