Skip to content

About

DEEPX NPU runtime and Linux driver for edge AI inference: DX-RT, firmware, DX-APP C++/Python examples and DX-Stream GStreamer plugins. Runs .dxnn models on DX-M1 M.2 accelerators.

Topics

Resources

Stars

24 stars

Watchers

1 watching

Forks

Latest commit

 

History

123 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

DEEPX DX-Runtime

DXNN® - DEEPX NPU SDK (DX-AS: DEEPX All-Suite)

DX-AS (DEEPX All-Suite) is an integrated environment of frameworks and tools that enables inference and compilation of AI models using DEEPX devices. Users can build the integrated environment by installing individual tools, but DX-AS maintains optimal compatibility by aligning the versions of the individual tools.

DXNN SDK Full Architecture DXNN SDK Simple Architecture

AI Model Runtime Environment (Deployment Platform)

Purpose

  • Must be installed on the Target system where the DEEPX M1 M.2 module is attached and the DEEPX AI model (.dxnn) will be executed.

Core Components

  • DX-RT & DX-FW & NPU Driver: Foundational software for NPU control
  • DX-APP: C++ and Python examples to jump-start your projects
  • DX-Stream: GStreamer integration for seamless multimedia pipelines

Flexibility & Support

  • OS: Compatible with Debian-based Linux (Ubuntu 26.04/24.04/22.04/20.04(LTS), Debian 13/12)
  • Architecture: Supports both x86_64 and arm64

Easy Installation

  • Our script automates the entire process
  • One-time reboot is required after installation to finalize the NPU Driver setup

You can install dx-runtime by following the instructions at this LINK.


Quick Guide (Install and Run)

DX-Runtime includes source code for each module. The repositories are managed as Git submodules(dx_rt_npu_linux_driver, dx_fw, dx_rt, dx_app, and dx_stream) under ./dx-runtime.

Local Installation

DX-Runtime supports installation in local environments.

You can install DX-Runtime by following the instructions at this Link

APT Repository Installation

The NPU driver and DX-RT are also published as Debian packages in the DEEPX APT repository (Ubuntu / Debian, amd64 / arm64), so you can install and upgrade them with apt.

Verify the signing key before trusting it: gpg --show-keys must print the fingerprint 3A20CC853C64AE328D0F58CFD816AAC6689DBDEA. If it does not match, stop — do not install the key.

wget -O deepx-archive-keyring.asc https://apt.releases.deepx.ai/gpg
gpg --show-keys deepx-archive-keyring.asc   # fingerprint must be 3A20CC853C64AE328D0F58CFD816AAC6689DBDEA
sudo gpg --dearmor -o /usr/share/keyrings/deepx-archive-keyring.gpg deepx-archive-keyring.asc
rm deepx-archive-keyring.asc

echo "deb [arch=$(dpkg --print-architecture) signed-by=/usr/share/keyrings/deepx-archive-keyring.gpg] https://apt.releases.deepx.ai $(. /etc/os-release && echo "${UBUNTU_CODENAME:-$VERSION_CODENAME}") main" | sudo tee /etc/apt/sources.list.d/deepx.list

sudo apt update
sudo apt install dxrt-driver-dkms   # NPU kernel driver, built for your kernel via DKMS
sudo apt install libdxrt-bin        # DX-RT library, dxcli and tools
  • The dx_engine Python wheels matching the installed libdxrt-bin are staged in /usr/share/libdxrt-bin/python; install the one for your Python version into your virtualenv.
  • Firmware is not part of these packages; update it by running ./install.sh --target=dx_fw from the dx-runtime root. A reboot after installing the driver is recommended.
  • dx_app and dx_stream are not covered by this route; use the Local Installation guide for them.

See the APT Repository Install section for details.

Docker Installation

DX-Runtime supports installation in docker environments.

You can install DX-Runtime by following the instructions at this Link

One-Line Installation (Runtime-Only)

For a quick setup of just the NPU driver, DX-RT, and DX-FW — without cloning this repository — run:

curl -fsSL https://raw.githubusercontent.com/DEEPX-AI/dx-runtime/main/oneline-install.sh | sh

This installs dx_rt_npu_linux_driver, dx_rt, and dx_fw from the prebuilt binaries on each component's main branch, so it always tracks the newest published build. It does not cover dx_app or dx_stream — for those, use the Local or Docker Installation above.

Pin a component instead of tracking main with DX_RT_VERSION, DX_DRIVER_VERSION, or DX_FW_VERSION, e.g. curl -fsSL ... | DX_RT_VERSION=3.5.0 sh. DX_FW_VERSION covers M1, M1M and H1 together — the three are released as one firmware set, and H1 ships inside the M1 version directory. Because the versions are resolved at run time, the artifacts are not checksum-verified, and the three components follow their own branches independently — a run between releases may install a combination that has not been validated together. Use the version pins or install.sh when you need a known-good set.

  • If no NPU device is detected, the firmware update step is skipped with a warning; rerun the same command once the device is available.

To remove it again without cloning the repository:

curl -fsSL https://raw.githubusercontent.com/DEEPX-AI/dx-runtime/main/oneline-uninstall.sh | sh

This purges the dxrt-driver-dkms and libdxrt-bin packages, which is what both install routes produce, so it removes an install.sh install just as well. Firmware already flashed to the device is not reverted — there is no uninstall path for it — and the dx_engine Python wheel is user-managed, so libdxrt-bin leaves it in place and prints how to remove it. For dx_app and dx_stream, which the one-liner never installs, use the repository's uninstall.sh.


Create User Manual

Install Python Dependencies

To install the necessary Python packages, run the following command:

pip install mkdocs mkdocs-material mkdocs-video pymdown-extensions mkdocs-with-pdf 

Generate Documentation (HTML and PDF)

To generate the user guide as both HTML and PDF files, execute the following command:

mkdocs build

This will create:

  • HTML documentation in the docs/ folder - open docs/index.html in your web browser
  • PDF file: DEEPX_[sub-module]_UM_[version]_[release_date].pdf in the root directory

About

DEEPX NPU runtime and Linux driver for edge AI inference: DX-RT, firmware, DX-APP C++/Python examples and DX-Stream GStreamer plugins. Runs .dxnn models on DX-M1 M.2 accelerators.

Topics

Resources

Stars

24 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages