Quick start

This guide explains how to get started with Radxa AIcore DX-M1 / DX-M1M.

Product photos

DX-M1

AIcore DX-M1 module (front)

AIcore DX-M1 module (back)

DX-M1M

AIcore DX-M1M module (front)

AIcore DX-M1M module (back)

Prerequisites

Prepare the following hardware to use AIcore DX-M1 / DX-M1M:

• AI accelerator module: Radxa AIcore DX-M1 / DX-M1M
• Host board kit: A compatible host board + power adapter + boot media (such as a microSD card)

Requirements

• AIcore DX-M1

The host board must provide an available M.2 M Key slot, supporting PCIe Gen 3.0

• AIcore DX-M1M

The host board must provide an available M.2 M Key or B Key slot, supporting PCIe Gen 3.0

Compatibility

AIcore DX-M1 / DX-M1M has been verified to work with multiple Radxa single-board computers (SBCs).

Verified Radxa-compatible boards

- ROCK 5A
- ROCK 5B
- ROCK 5B+
- ROCK 5 ITX

Hardware installation

Follow the Hardware installation guide to install AIcore DX-M1 / DX-M1M.

Note

For most host boards, installing an M.2 M Key add-on device follows the same general procedure. You can refer to the Hardware installation guide.

Verify hardware detection

After installing AIcore DX-M1 / DX-M1M into the host board's M.2 M Key slot and booting the OS, you can run lspci to verify whether the device is detected:

Device

lspci

If AIcore DX-M1 / DX-M1M is detected, you should see output similar to the following:

0000:01:00.0 Processing accelerators: DEEPX Co., Ltd. DX_M1A

Environment setup

On Linux (Ubuntu, Debian), refer to DX-NPU-Driver installation and DX-RT runtime installation to install the required driver and runtime environment for AIcore DX-M1 / DX-M1M.

Example applications

After completing the steps above, you can follow DX-APP examples to run AI acceleration and inference with AIcore DX-M1 / DX-M1M.