VeloCT

Public release for VeloCT / H-Houdini. This project focuses on invariant learning for hardware verification. For a detailed overview, refer to our ASPLOS'25 publication.

Clone the Repository

Before cloning, ensure that you have Git LFS installed:

# Install Git LFS
# Ubuntu/Debian
sudo apt install git-lfs

# macOS
brew install git-lfs

git lfs install

Then, clone the repository:

git clone https://github.com/FPSG-UIUC/veloct.git
cd veloct

Setup

System-wide Dependencies

1. Install and run Redis:
   Redis is required for distributed processing. Install and start it using:

   # Ubuntu/Debian
   sudo apt update && sudo apt install redis
   sudo systemctl enable redis-server --now
   
   # macOS
   brew install redis
   brew services start redis

2. Install and run MongoDB (optional, required for debugging):

   # Ubuntu/Debian
   sudo apt install mongodb
   sudo systemctl enable mongod --now
   
   # macOS
   brew tap mongodb/brew
   brew install mongodb-community@6.0
   brew services start mongodb-community@6.0

   NOTE: Ensure MongoDB is running without errors:

   sudo systemctl status mongod  # Ubuntu
   brew services list             # macOS

Python Dependencies

This project uses Poetry for dependency management. Ensure you have it installed:

# Ubuntu/Debian
curl -sSL https://install.python-poetry.org | python3 -

# macOS
brew install poetry

Then, create a new virtual environment and install dependencies:

poetry install

Running the Application

Default Run Command For Rocketchip

To start invariant learning:

python3 -m learning.learn_inv_distributed learn-invariant

Running on BOOM

Current targets: smallboom,mediumboom,largeboom,megaboom

• First, checkout the boom branch: git checkout boom
• Next, prepare target: ./prepare_target.sh <target>; The target has to be one of above target strings.
• Run the same command as as before

Managing Logging Output

By default, logging is verbose, which may slow down invariant learning. To reduce log verbosity, set the log level to INFO:

LOGURU_LEVEL="INFO" python3 -m learning.learn_inv_distributed learn-invariant

Retrieving Results

To print the learned invariant:

python3 -m learning.learn_inv_distributed get-invariant

For additional options:

python3 -m learning.learn_inv_distributed --help