Logic Synthesis Optimization with Predictive Self-Supervision via Causal Transformers

TL;DR

LSOformer addresses QoR prediction in Logic Synthesis Optimization by using a decoder-only transformer that fuses AIG representations with optimization recipes via cross-attention, guided by predictive self-supervision of QoR trajectories. The approach combines a graph encoder with level-wise pooling, heuristic tokenization, and a transformer decoder to produce step-wise QoR predictions, optimized with a joint SSL and regression objective. Empirical results on three datasets show notable improvements over baselines in both delay and area QoR, with ablations confirming the benefits of SSL and the fusion architecture. The method offers practical impact by enabling faster QoR estimation to guide synthesis choices under data-scarce conditions, advancing ML-assisted EDA pipelines.

Abstract

Contemporary hardware design benefits from the abstraction provided by high-level logic gates, streamlining the implementation of logic circuits. Logic Synthesis Optimization (LSO) operates at one level of abstraction within the Electronic Design Automation (EDA) workflow, targeting improvements in logic circuits with respect to performance metrics such as size and speed in the final layout. Recent trends in the field show a growing interest in leveraging Machine Learning (ML) for EDA, notably through ML-guided logic synthesis utilizing policy-based Reinforcement Learning (RL) methods.Despite these advancements, existing models face challenges such as overfitting and limited generalization, attributed to constrained public circuits and the expressiveness limitations of graph encoders. To address these hurdles, and tackle data scarcity issues, we introduce LSOformer, a novel approach harnessing Autoregressive transformer models and predictive SSL to predict the trajectory of Quality of Results (QoR). LSOformer integrates cross-attention modules to merge insights from circuit graphs and optimization sequences, thereby enhancing prediction accuracy for QoR metrics. Experimental studies validate the effectiveness of LSOformer, showcasing its superior performance over baseline architectures in QoR prediction tasks, where it achieves improvements of 5.74%, 4.35%, and 17.06% on the EPFL, OABCD, and proprietary circuits datasets, respectively, in inductive setup.

Figures (2)

• Figure 1: Schematic overview of the LSOformer architecture during training and inference modes.
• Figure 2: Mean and circuit-wise performance comparison across test circuits for Area (a) and Delay (b).