The Xilinx Certified Ubuntu 20.04 LTS for Xilinx Devices image is an official Ubuntu image with certified hardware support for select Xilinx evaluation boards. This page details how to boot and use the official desktop environment image released by Canonical for Xilinx ZCU102, ZCU104, and ZCU106 evaluation boards. You can read more about certified Ubuntu hardware here.
(Optional) USB hub for peripherals (ex: keyboard, mouse)
ZCU106 (Rev 1.0+)
8GB+ SD Card
(Optional) USB hub for peripherals (ex: keyboard, mouse)
Note: The ZCU104 & ZCU106 VCU ROI TRD demonstrations additionally require the following accessories: HDMI monitor, HDMI source device (ex: DVD player), and HDMI cables. HDMI is not used on the ZCU102 design.
Installing the Image
Once you have downloaded the Certified Ubuntu for Xilinx Devices image fromhttp://ubuntu.com/download/xilinx, you can write it to your SD card using any disk writing tool such as Balena Etcher, Win32 Disk Imager, or dd.
The screenshots below show how to write the disk image with Win32 Disk Imager.
If you choose to boot without the optional accessories, you can monitor the boot process via the USB UART terminal output. Once you see the zynqmp login: prompt you can log into the system.
The default Ubuntu 20.04 LTS terminal login prompt
The default login credentials are:
The standard system policy requires you to change the password after the first time you log in with the default credentials.
If you login very quickly the first boot, you may find that the password does not have to be changed. In this case, you will be required to update it the next time you login or use sudo.
The default Ubuntu 20.04 LTS terminal
Logging in via the Graphical Desktop
If you are using the graphical interface via a monitor, the login process is similar:
the default Ubuntu 20.04 LTS login greeter screen
the default Ubuntu 20.04 LTS GNOME 3 desktop
Note: All systems using the Certified Ubuntu for Xilinx Devices image will initially boot with the graphical resolution of 1920x1080. 4K (3840x2160 or higher) resolutions are possible but Xilinx recommends using the 1920x1080 resolution for the most robust desktop experience.
The Ubuntu Desktop configuration used for this image is ubuntu-desktop-minimal. This configuration does not include Ubuntu Software by default. While you can always install Debian packages and snaps from the command line, if you would like to install the Ubuntu Software application, you can run the following command in a terminal:
sudo snap install snap-store
The instructions below assume the system is connected to the Internet. If your system is not connected to the Internet please connect the Ethernet cable and configure the system for Internet access before continuing.
If your network is configured with a DHCP server, the Certified Ubuntu for Xilinx Devices system will automatically obtain an IP address. If a static address is required, you can use the ip command from the command line to configure the network interface. For more information, please see https://ubuntu.com/server/docs/network-configuration
For those users that prefer to use the ifconfig tools to configure your network interface, you can install the net-tools package. These tools are no longer included as part of the standard Ubuntu 20.04 root filesystem. To install them, you can use the following command: $sudo apt install net-tools
If you are using the Gnome 3 graphical interface, you can configure networking via the Network Settings options in the GUI.
The GNOME 3 Network Settings GUI
Set up the Xilinx Development & Demonstration Environment for Ubuntu 20.04 LTS
This step requires Internet access. If you haven’t already set up Internet access, please follow the instructions in the Network Setup section above
Before diving into the demos, the Certified Ubuntu on Xilinx Devices environment needs to have some additional resources downloaded and installed. The primary utility for switching between configurations and managing the system is called xlnx-config. The xlnx-config utility is deployed as a Snapcraft-compatible snap.
Install the custom Xilinx gstreamer
Downloads a Vitis AI Model
To install the xlnx-config snap, execute the following command from a terminal:
$ sudo snap install xlnx-config --classic
After installing the xlnx-config snap, you can use it to set up the remainder of the system by executing the following command in a terminal:
NOTE: The Xilinx team updates the xlnx-config snap with new features over time. It is important to keep it up-to-date.
The heart of the Xilinx demonstrations is the Xilinx Deep Learning Processing Unit (DPU). In addition to setting up the system, the xlnx-config snap provides the DPU via dpu.xclbin to other snaps that depend on it. The xlnx-config snap also provides infrastructure which allows users to package up other hardware platforms to enable other custom development and demonstration workflows. For more information about creating and installing custom platforms via xlnx-config, see the Usage Example section on the xlnx-config developer page.
Install the Custom Xilinx gstreamer
By default, the Ubuntu 20.04 LTS root filesystem includes the open source upstream version of gstreamer that is packaged with Ubuntu 20.04 LTS. In order to take advantage of the unique features provided by Xilinx multimedia IP, xlnx-config.sysinit installs a customized Xilinx-specific version of gstreamer with specialized extensions and plugins.
Download a Vitis AI Model
During xlnx-config.sysinit, the densebox_640_360 model is downloaded from the Xilinx Model Zoo to facilitate running of the VCU ROI TRD Demo.
Overview of the Out-of-the-Box Demos
Each evaluation board boot image contains a default bitstream & device tree combination that is configured during the first boot. Each platform includes the Xilinx Deep Learning Processing Unit (DPU) in the Zynq UltraScale+ Programmable Logic (PL). These platforms also include a software stack (XRT + Vitis AI) required to take advantage of it. All of the out-of-the-box images are based on Xilinx 2020.2 tools, IP, drivers, and firmware. The default platforms for the three boards are:
Each of the Out-of-the-Box demo designs includes the Xilinx DPU for machine learning inference. The ZCU104 & ZCU106 designs include a 4096x1 configuration. The ZCU102 includes 4096x3 configuration. Each DPU configuration has an identical feature set and they are compatible with the pre-built models in the Xilinx Model Zoo that target the ZCU10x boards.
Running the Vitis AI Library Samples Snap
The xlnx-vai-lib-samples snap includes a pre-built subset of the Vitis AI Library sample applications available as source in the Vitis AI Github repository. After installing the snap, users can experiment with twenty-seven different Vitis AI samples and over 90 different pre-built models.
Using the ZCU104 & ZCU106 VCU Region of Interest (ROI) TRD Demo
This step requires an HDMI video source, HDMI display, and HDMI cables
The primary goal of the VCU ROI design is to demonstrate the use of the Xilinx Deep learning Processor Unit (DPU) block for extracting the Region of Interest (ROI) data from input video frames. This information is then used to perform ROI-based encoding using the Video Codec Unit (VCU) encoder hard block present in Zynq UltraScale+ EV devices.
Note: Be sure to delimit quotation marks around the LINUX_KERNEL_CMDLINE values. This is so that multiple arguments can be passed at the same time. If only one argument is being specified then they can be omitted.
Note: If command line arguments are being used which include forward slashes ( / ) such as those found in filesystem paths, they must be delimited twice.
Example: sudo sh -c 'echo "LINUX_KERNEL_CMDLINE=\"cma=256M\ root=\\/dev\\/sda1" > /etc/default/u-boot-xlnx'
You must reboot the system for the changes to take place.
Installing & Using the Xilinx PPA
Xilinx has a Personal Package Archive (PPA) hosted by Canonical Launchpad at https://launchpad.net/~ubuntu-xilinx/+archive/ubuntu/updates . This PPA contains packages for some Xilinx-specific binaries that are not available in other upstream repositories. The standard method for installing PPA’s in Ubuntu desktop environment is documented here. To install the PPA via the command line, issue the following commands:
If you used the xlnx-config.sysinit to setup the system after first boot, the PPA has already been installed.
Libraries required to support XRT on Arm platforms
Checking the Xilinx Kernel Configuration
If you would like to check the configuration of the current Ubuntu kernel, you can find the file in /proc/config.gz. You can output the contents of this file using the zgrep command:
$ zgrep /proc/config.gz
Special Considerations for the ZCU104 Evaluation Board
The ZCU104 evaluation board has less physical memory than the other supported boards. This difference means that resolutions higher than 1080p are not supported on the GNOME Desktop. Switching to a resolution lower than 1080p (e.g. 1280x720) can create a more responsive desktop experience. Full-screen video playback on the desktop should also be avoided to reduce the chance of running into out-of-memory performance issues.
Here are some recommendations for using the ZCU104 with this release:
Before running xlnx-config.sysinit from the Gnome Desktop:
Ensure that nothing else is running in the background.
Change the desktop resolution to 1280x720.
Do not update the desktop background
If you want the best desktop experience but are not interested in using the VCU, you can reduce the amount of contiguous memory(CMA) that is allocated by changing the kernel bootargs. 256M is recommended. Please refer to the Changing the Kernel bootargs Used By U-Boot section.
If the desktop gets into an unstable state, you can try restarting it from the console with the following command:
sudo systemctl restart gdm
If you’re working from the serial port console or through ssh, you can shutdown the desktop to maximize available memory:
sudo systemctl stop gdm
DisplayPort to HDMI Adapters
Historically, DisplayPort to HDMI adapters have been problematic when used with the ZCU10x boards. However, there are some that work consistently. The table below lists ones that are known to work as well as some that did not work.