ShortCircuit: AlphaZero-Driven Circuit Design

TL;DR

ShortCircuit is a transformer-based approach that combines supervised and reinforcement learning, inspired by AlphaZero, to efficiently generate optimized Boolean circuits from truth tables, surpassing traditional methods.

Contribution

It introduces a novel AlphaZero-inspired framework for logic circuit synthesis, leveraging structural properties and a two-phase learning process for improved generalization and performance.

Findings

Generates AIGs for 98% of test truth tables

Outperforms ABC by 18.62% in circuit size

Effective in exploring large search spaces

Abstract

Chip design relies heavily on generating Boolean circuits, such as AND-Inverter Graphs (AIGs), from functional descriptions like truth tables. This generation operation is a key process in logic synthesis, a primary chip design stage. While recent advances in deep learning have aimed to accelerate circuit design, these efforts have mostly focused on tasks other than synthesis, and traditional heuristic methods have plateaued. In this paper, we introduce ShortCircuit, a novel transformer-based architecture that leverages the structural properties of AIGs and performs efficient space exploration. Contrary to prior approaches attempting end-to-end generation of logic circuits using deep networks, ShortCircuit employs a two-phase process combining supervised with reinforcement learning to enhance generalization to unseen truth tables. We also propose an AlphaZero variant to handle the double exponentially large state space and the reward sparsity, enabling the discovery of near-optimal designs. To evaluate the generative performance of our model , we extract 500 truth tables from a set of 20 real-world circuits. ShortCircuit successfully generates AIGs for $98\%$ of the 8-input test truth tables, and outperforms the state-of-the-art logic synthesis tool, ABC, by $18.62\%$ in terms of circuits size.