PYNQ Composable Pipeline & Partial Programming

Outline of the lecture

• PYNQ
  1. PYNQ Introduction & Background
  2. PYNQ Overlays
  3. PYNQ Libraries
  4. Overlay Design Methodology
• PYNQ Composable Overlays
  1. A Composable Video Pipeline
  2. Composable Overlay Methodology
  3. Default Paths
• Tutorial
  1. Composable Video Pipeline

A Composable Video Pipeline

To demonstrate the benefits of the composable overlay, we reference a composable video pipeline from offical https://github.com/Xilinx/PYNQ_Composable_Pipeline

Setup Juypter notebook examples of Composable Video Pipeline

Preparation:

• PYNQ-Z2 board with pynq 2.7up
• HDMI displayer
• HDMI input supports 720P (tutorial 02~06)
• USB Webcam supports 720P (tutorial 08)

Open a Jupyter Lab terminal and install Composable Video Pipeline

$ root@pynq:/# cd /home/xilinx/
$ root@pynq:/home/xilinx# git clone https://github.com/Xilinx/PYNQ_Composable_Pipeline
$ root@pynq:/home/xilinx# python3 -m pip install PYNQ_Composable_Pipeline/ --no-build-isolation
$ root@pynq:/home/xilinx# pynq-get-notebooks pynq_composable -p $PYNQ_JUPYTER_NOTEBOOKS

Build Composable Video Pipeline sources on Vivado server

Build Environment:

• Ubuntu 20.04 and Vivado 2020.2
• Pre-installed libc6-dev-i386 package
• Pre-installed PYNQ-Z2 board files

Make Composable Video Pipeline sources

$ git clone https://github.com/Xilinx/PYNQ_Composable_Pipeline --recursive
$ cd PYNQ_Composable_Pipeline/boards/Pynq-Z2/
$ make