Skip to content

Install Vivado 2022.2

You need Vivado only to change the FPGA design. It is about 50 GB installed, and the Zynq-7020 is covered by the free WebPACK licence. Not changing the FPGA? Build without Vivado.

The installer will not run on a host newer than Ubuntu 22.04, so the container runs it and writes to the host. Download the installer yourself first: AMD put it behind an account login.

# run from: the repo root
# AMD put the installer behind an account login, so download it yourself first:
#   https://www.xilinx.com/support/download.html  ->  Vivado 2022.2  ->  Linux Self Extracting Web Installer

# Rootless podman maps the container's root to YOUR user, so the target must be
# yours: a root-owned /tools/Xilinx cannot be written even from "root" inside.
sudo mkdir -p /tools/Xilinx && sudo chown "$USER" /tools/Xilinx

./devkit container install ~/Downloads/Xilinx_Unified_2022.2_1014_8888_Lin64.bin

The web installer needs your AMD account and downloads the content itself; budget an hour.

  1. Create an account at xilinx.com and open the 2022.2 downloads page.
  2. Download the Vitis unified installer for Linux (it offers both products; this project uses only Vivado).

  3. Run it:

    # run from: the directory holding the installer, on your host
chmod +x Xilinx_Unified_2022.2_*.bin && ./Xilinx_Unified_2022.2_*.bin

Either way, answer the installer the same:

It asks Choose
product Vivado, edition Vivado ML Standard
device families only Zynq-7000 (~130 GB down to ~30 GB)
path keep the default /tools/Xilinx, which tools/env-vivado.sh points at

You should see: ./devkit doctor --target factory (or ./devkit container doctor --target factory) find Vivado.

Always source tools/env-vivado.sh, never Vivado's own settings64.sh

The script supplies the old libraries Vivado 2022.2 needs (why).

The installer says \"Extraction failed\"?

Usually false: the container's install step checks for an executable xsetup instead. When it is real, and what the install mounts: installing Vivado in the first place.

Next: build with Vivado in a container, then add your own logic to the FPGA.