Radxa Launches AICore DX-M1M Edge AI Accelerator Module with 25TOPS Computing Power

Radxa recently launched the AICore DX-M1M compact edge AI accelerator module, based on DEEPX's DX-M1M chip, targeting edge inference scenarios for embedded and industrial systems.

The module uses M.2 2242 B+M Key form factor, PCIe Gen3×2 interface, onboard 1GB LPDDR4X memory and 1Gbit storage, delivering 25TOPS INT8 computing power with only 3W power consumption.

It supports x86 and Arm platforms, Windows, Linux Ubuntu distributions, Docker environments, and is compatible with mainstream AI frameworks like TensorFlow and ONNX via the DX-COM compiler.