CAVEAT

Context-Aware Verification, Emulation, and Training

Scope

Verification using vendor-specific as well as independent simulators and frameworks is a fundamental and well established step to successful digital design. Similarly, software-defined configuration and operation of hardware has significantly improved accessibility of technology. In fact, modern software-defined experiment design allows scientists and cross-disciplinary users to propose novel modes of operation within hours or even minutes.

This vendor-agnostic framework decouples application development, design verification, and training from the immediate availability of sensitive, unwieldy, and costly hardware installations. Moreover, it facilitates feasibility studies prior to investments as well as reproducibility of research results.

The framework is designed with three distinct needs of academic and development laboratories in mind:

1. streamlining context-aware verification for supervised operation,
2. hardware emulation for ease of application software development, and
3. an environment for experiential training of prospective users.

It distinguishes two distinct aspects of verification:

1. static (topologic), and
2. dynamic (algorithmic).

For basic examples on how to do this, please see slides presented at FPGA Conference Europe 2025.

Dependencies

(see requirements.txt)

Installation (inside virtual environment)

sudo apt install python-venv

activate virtual env

python3 -m venv caveatenv
source caveatenv/bin/activate

Then ensure build is installed

python3 -m pip install build

Install locally

In the caveat top level directory where pyproject.toml lives, run:

pip install -e .

To test the installed code, run:

cd example
./run_tests.py

The test test_loopback_throughput will execute, and the results are placed in the subdirectory build/results/.

Recommended Tools (optional)

• python 3.11+
• pip
• venv
• pytest-reporter-html-dots for advanced html reporting
• pytest-xdist for test parallelization (or pytest-parallel)