27 March 2025 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
This article introduces OpenLS-DGF, an adaptive logic synthesis dataset generation framework, to enhance machine-learning (ML) applications within the logic synthesis process. Previous dataset generation flows were tailored for specific tasks or lacked integrated ML capabilities. While OpenLS-DGF supports various ML tasks by encapsulating the three fundamental steps of logic synthesis: 1) Boolean representation; 2) logic optimization; and 3) technology mapping. It preserves the original information in both Verilog and ML-friendly GraphML formats. The Verilog files offer semi-customizable capabilities, enabling researchers to insert additional steps and incrementally refine the generated dataset. Furthermore, OpenLS-DGF includes an adaptive circuit engine that facilitates the final dataset management and downstream tasks. The generated OpenLS-D-v1 dataset comprises 46 combinational designs from established benchmarks, totaling over 966 000 Boolean circuits. OpenLS-D-v1 supports integrating new data features, making it more versatile for new tasks. This article demonstrates the versatility of OpenLS-D-v1 through four distinct downstream tasks: circuit classification, circuit ranking, quality of results (QoR) prediction, and probability prediction. Each task is chosen to represent essential steps of logic synthesis, and the experimental results show the generated dataset from OpenLS-DGF achieves prominent diversity and applicability. The source code and datasets are available at https://github.com/Logic-Factory/ACE/blob/master/OpenLS-DGF.
31 December 2024 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Technology mapping is the pivotal synthesis step that translates abstract logical models into technology-dependent implementations using the designated library, e.g., standard cells for ASICs. The efficient solutions heavily rely on the gate selection guided by estimated delay. However, estimating these delays is sophisticated due to the absence of actual interconnect load and transition time during the mapping. In this article, we revisit the difficulties of the delay-driven mapping problem and explore three key insights to address these. Inspired by the insights, we first design a structure-aware load-slew model that integrates input transitions and output loads for gate delay estimations. Benefiting from the model, we propose a delay-iterative framework that progressively reduces the overall circuit delay by further aligning library characteristics with logical network structures. Finally, experiments with 130 nm and 7 nm libraries show its superiority, which averagely reduces circuit delay by 10% with nonlinear delay model, and 6% in delay after P&R, as compared to ABC.
21 March 2026 in ACM Transactions on Design Automation of Electronic Systems
Logic Optimization (LO) is a critical stage in the chip design process, focused on improving the Quality of Results (QoR) by optimizing circuit designs to minimize area and delay. During logic optimization, evaluating the QoR after each iteration requires completing logic optimization and technology mapping. The evaluation process is highly time-consuming, restricting the number of optimization iterations possible within a given time. To address this, the AI-aided logic optimization framework (AiLO) is developed to explore more optimization operator sequences (recipes). AiLO framework consists of two core components: AI-based metric evaluation and optimization exploration. To achieve accurate evaluation, different prediction models can be integrated. A multi-scale cross-attention Transformer (CrossLO) is introduced to simulate the optimization structure of recipes across circuit at various scales to enhance the prediction accuracy. Moreover, the AI evaluation module can effectively maintain the recipe ranking, even when prediction accuracy is biased. The logic optimization exploration algorithm integrated with CrossLO (AI evaluation) shows an average improvement of 14.75% over the initial version. NSGA-II (optimization module) integrated with CrossLO achieves a significant lead over other algorithms in the same time. In addition, the AiLO framework continues to grow with the performance of the two components, demonstrating strong adaptability and flexibility.
28 January 2026 in ACM Transactions on Design Automation of Electronic Systems
The increasing complexity of digital circuits and the limitations of heuristic methods have led to growing interest in applying Machine Learning (ML) to Logic Synthesis (LS). ML provides a promising paradigm shift by implementing automated, scalable, and data-driven optimization strategies. This survey provides a comprehensive overview of the latest studies on ML approaches in LS, offering a deep understanding of the fundamental ML methods, and analyzing their strengths and limitations through systematic comparisons. We categorize existing works into two main types: ML-Assistance methods aiming at predicting performance metrics and reducing the cost of traditional simulations, and ML-Agent methods directly replacing heuristic processes in the LS flow. We further analyze ML methods and applications in different LS stages, including Boolean circuit generation, Boolean circuit analysis, logic optimization, and technology mapping, showing great achievements and improvement in exploring non-linear design spaces and discovering new optimization strategies. Finally, we discuss the challenges and limitations in the current situation and further provide a vision of future directions in LS.
1 July 2026 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
1 November 2026 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Graph neural networks have emerged as a powerful tool for learning circuit representation but largely ignore the dynamic nature of logic optimization (LO), leading to a severe “domain gap” where embedding quality degrades after heuristic transformations. To address this, we propose CircuitFlow, a framework learning representations for both circuits and optimization algorithms. CircuitFlow is distinguished by three features: (1) it employs circuit motifs and DAGTrans for robust encoding; (2) it utilizes a dual-component architecture where OPTrans models optimization operations via Transformers; and (3) it seamlessly integrates into LO pipelines to predict optimized circuit statistics.Experimental results demonstrate that CircuitFlow reduces physical metric prediction loss by up to 72% compared to baselines. Crucially, it effectively bridges the domain gap, bringing the generalization error on optimized circuits down from over 65% to within 20%. Furthermore, integrating CircuitFlow into the LO flow yields an average QoR improvement of 3.97% alongside a 3.11× speedup. Finally, we leverage CircuitFlow to introduce orientational logic optimization, a new paradigm guiding synthesis toward target structures, which reduces the runtime of Combinational Equivalence Checking (CEC) by 76% on average.