Installation#

Supported Configurations#

The Ryzen AI Software Platform supports AMD Ryzen 7040U, 7040HS series mobile processors with Windows 11 OS.

Prepare Client Device#

Download the IPU Driver and install it by following these steps:

  1. Extract the downloaded zip file.

  2. Open a terminal in administrator mode and execute the .\amd_install_kipudrv.bat bat file.

Ensure that the IPU driver is installed from Device Manager -> System Devices -> AMD IPU Device as shown in the following image.

_images/driver_0_9.png


Note: If you see the “Windows could not verify the digital signature of this driver” error message, follow this tutorial to disable signature checking.

To enable the development and deployment of IPU-based inference on the client device, you must have the following software installed, along with their minimum versions.

Dependencies

Version Requirement

Visual Studio

2019

cmake

version >= 3.26

python

version >= 3.9 (3.9.13 64bit recommended)

Anaconda or Miniconda

Latest version

AMD IPU driver

>= 10.105.5.42



Installation Steps#

Download the setup package ryzen-ai-sw-0.9.tar.gz and extract.

https://account.amd.com/en/forms/downloads/ryzen-ai-software-platform-xef.html?filename=ryzen-ai-sw-0.9.zip

Before running the installation script ensure all the prerequisites outlined above have been met and the Windows PATH variable is properly set for each component. Make sure to set the PATH variable using “Environment Variables” window from the “System Properties”. For example, Anaconda requires following paths to set the PATH variable path\to\anaconda3\, path\to\anaconda3\Scripts\, path\to\anaconda3\Lib\bin\.

Open the Windows Command Prompt and run the below command

.\install.bat

The installation script, install.bat does the following

  • Creates a conda environment

  • Installs Vitis AI ONNX Quantizer

  • Installs ONNX Runtime

  • Installs Vitis AI Execution Provider

  • Sets the environment variable to specify 1x4.xclbin binary

  • Prints the name of the conda environment before exiting

Thereafter you can activate the created conda environment to get the work environment.

To provide a specific name of the conda work environment run the install.bat as shown below

.\install.bat -env <env name>
  • Instead of the bundled installation process using install.bat, if you choose to install each component manually please refer to Manual Installation page.

  • If you require to install Vitis AI PyTorch/TensorFlow or Microsoft Olive Quantizer, please refer to Alternate Quantizer Setup page.



Runtime Environment Setup#

Runtime IPU Binary Selection#

The IPU binaries are located inside the setup package. Selecting an IPU binary is a required step every time the application is run from a new terminal. For example, IPU binary 1x4.xclbin is selected as below

set XLNX_VART_FIRMWARE=C:\path\to\ryzen-ai-sw-xx\ryzen-ai-sw-xx\voe-xx-win_amd64\1x4.xclbin

Ryzen AI Software platform provides multiple IPU binaries using different configurations on the IPU device. Refer to the Runtime Setup page for more details on IPU binaries.

Runtime Configuration File#

The Execution Provider setup package contains the Vitis AI Execution Provider runtime configuration file vaip_config.json. This file is required when configuring Vitis AI Execution Provider (VAI EP) inside the ONNX Runtime code.