Vitis AI v1.4

Introduction

This directory contains instructions for running DPUCZDX8G on Zynq Ultrascale+ MPSoC platforms. DPUCZDX8G is a configurable computation engine dedicated for convolutional neural networks. It includes a set of highly optimized instructions, and supports most convolutional neural networks, such as VGG, ResNet, GoogleNet, YOLO, SSD, MobileNet, FPN, and others. With Vitis-AI, Xilinx has integrated all the edge and cloud solutions under a unified API and toolset.

Step1: Setup cross-compiler

1. Run the following command to install cross-compilation system environment.

Please install it on your local host linux system, not in the docker system.

./host_cross_compiler_setup.sh

Note that the Cross Compiler will be installed in ~/petalinux_sdk_2021.1 by default. For the kv260, use host_cross_compiler_setup_2020.2.sh to install the cross-compiler.

2. When the installation is complete, follow the prompts and execute the following command.

source ~/petalinux_sdk_2021.1/environment-setup-cortexa72-cortexa53-xilinx-linux

Note that if you close the current terminal, you need to re-execute the above instructions in the new terminal interface.

Step2: Setup the Target

To improve the user experience, the Vitis AI Runtime packages, VART samples, Vitis-AI-Library samples and models have been built into the board image. Therefore, user does not need to install Vitis AI Runtime packages and model package on the board separately. However, users can still install the model or Vitis AI Runtime on their own image or on the official image by following these steps.

1. Installing a Board Image.

   • Download the SD card system image files from the following links:

     ZCU102

     ZCU104

     KV260

     Note: For ZCU102/ZCU104, the version of the board image should be 2020.2 or above.
     For KV260, the version of the board image is 2020.2.
     If you use 2020.2 system, use the corresponding 2020.2 cross-compiler.

   • Use Etcher software to burn the image file onto the SD card.

   • Insert the SD card with the image into the destination board.

   • Plug in the power and boot the board using the serial port to operate on the system.

   • Set up the IP information of the board using the serial port.

   For the details, please refer to Setting Up the Evaluation Board

2. (Optional) Running zynqmp_dpu_optimize.sh to optimize the board setting.

   The script runs automatically after the board boots up with the official image. But you can also download the dpu_sw_optimize.tar.gz from here.

   cd ~/dpu_sw_optimize/zynqmp/
   ./zynqmp_dpu_optimize.sh

3. (Optional) How to install the Vitis AI for PetaLinux 2021.1
   There are two ways to install the dependent libraries of Vitis-AI. One is to rebuild the system by configuring PetaLinux and the other is to install the Vitis-AI online via dnf.

   • Build-Time
     To obtain the yocto recipes of VAI1.4 via petalinux-upgrade command, then rebuild the petalinux project . More details please refer to the PetaLinux Tools Documentation:Reference Guide(UG1144) Chapter 6 Upgrading the Workspace.
     For 2021.1 update1 release, run the following command and source the tool's setting script.

     rm <path to petalinux tool>/components/yocto/source/aarch64
     petalinux-upgrade -u 'http://petalinux.xilinx.com/sswreleases/rel-v2021/sdkupdate/2021.1_update1/' -p 'aarch64'
     source settings.sh
     Note that if you do not remove the old aarch64 file first, the upgrade may not be correct on some host system.
   • Run-Time
     Execute dnf install packagegroup-petalinux-vitisai to complete the installation on the target.
     Note: If you use this method, ensure that the board is connected to the Internet.

4. (Optional) How to update Vitis AI Runtime and install them separately.
   If you want to update the Vitis AI Runtime or install them to your custom board image, follow these steps.

   • Copy the following folder to the board using scp.

   scp -r mpsoc root@IP_OF_BOARD:~/
   • Log in to the board using ssh. You can also use the serial port to login.
   • Install the Vitis AI Runtime. Execute the following command.

   cd ~/mpsoc/VART
   bash target_vart_setup.sh

   Note that for the kv260, use target_vart_setup_2020.2.sh to install the VART.

5. (Optional) Download the model. For each model, there will be a yaml file which is used for describe all the details about the model. In the yaml, you will find the model's download links for different platforms. Please choose the corresponding model and download it. Click Xilinx AI Model Zoo to view all the models.

   • Take resnet50 of ZCU102 as an example.

     cd /workspace
     wget https://www.xilinx.com/bin/public/openDownload?filename=resnet50-zcu102_zcu104_kv260-r1.4.1.tar.gz -O resnet50-zcu102_zcu104_kv260-r1.4.1.tar.gz
   • Copy the downloaded file to the board using scp with the following command.

     scp resnet50-zcu102_zcu104_kv260-r1.4.1.tar.gz root@IP_OF_BOARD:~/
   • Log in to the board (using ssh or serial port) and install the model package.

     tar -xzvf resnet50-zcu102_zcu104_kv260-r1.4.1.tar.gz
     cp resnet50 /usr/share/vitis_ai_library/models -r

Step3: Run the Vitis AI Examples

1. Download the vitis_ai_runtime_r1.4.x_image_video.tar.gz from host to the target using scp with the following command.

   [Host]$scp vitis_ai_runtime_r1.4.*_image_video.tar.gz root@[IP_OF_BOARD]:~/

2. Unzip the vitis_ai_runtime_r1.4.x_image_video.tar.gz package on the target.

   cd ~
   tar -xzvf vitis_ai_runtime_r*1.4*_image_video.tar.gz -C Vitis-AI/demo/VART

3. Enter the directory of samples in the target board. Take resnet50 as an example.

   cd ~/Vitis-AI/demo/VART/resnet50

4. Run the example.

   ./resnet50 /usr/share/vitis_ai_library/models/resnet50/resnet50.xmodel

   For examples with video input, only webm and raw format are supported by default with the official system image. If you want to support video data in other formats, you need to install the relevant packages on the system.

References

• Vitis AI User Guide