Example of Python and PyTest powered workflow for a HDL simulation

Detailed description can be found on Russian at Habr.com

Repository structure

$ tree -a -I .git
.
├── .github
│   └── workflows # Github Actions
│       ├── icarus-test.yml # run all tests with Icarus Verilog on every push
│       └── modelsim-test.yml # run all tests with Modelsim on every push
├── .gitignore
├── LICENSE.txt
├── README.md
├── sim # simulation scripts
│   ├── conftest.py
│   ├── sim.py
│   └── test_sqrt.py
└── src # sources
    ├── beh # behavioral models
    │   └── sqrt.py
    ├── rtl # synthesizable HDL
    │   └── sqrt.v
    └── tb # testbenches
        └── tb_sqrt.sv

sim.py is the simulation core file, which contains wrappers for Icarus Verlog, Modelsim and Vivado simulator.

Description

Long story short, I simply wanted to replace my Bash scripts I usually use to simulate HDL. And actually I did it and earned even more that expected.

This is a simple example of how Python and Pytest can be combined to create a flexible and easy to support environment to simulate and test HDL.

Main idea is to wrap several simulators like Icarus Verilog, Modelsim or Vivado Simulator with one python class Simulator, add class CliArgs to parse command line arguments, implement some frequently used utility functions. And then simply use this things in any other Python script, so you will be able to:

• to run single selected test (e.g. test_sv) with GUI

./test_sqrt -t test_sv

• to run single test in batch mode without GUI

./test_sqrt -t test_sv -b

• to run single test in a specified simmulator

./test_sqrt -t test_sv -s modelsim

• to run single test with additional defines

./test_sqrt -t test_sv -d DIN_W=32 ITER_N=1000

• to run all tests available

pytest -v

• to run all tests in a parallel way (pytest-xdist must be installed)

pytest -v -n 4

• to run all tests in a different simulator

pytest -v --sim vivado

And as a bonus this workflow is easy to organize CI. Examples of Github Actions are in .github/workflows.