MappingEvolve: LLM-Driven Code Evolution for Technology Mapping

2026-04-29

AI Summary

Technology mapping remains a critical bottleneck in logic synthesis, with conventional approaches constrained in their optimization capabilities. This work presents the first integration of large language models into the evolution of core technology mapping algorithms, introducing an open-source framework that employs a hierarchical agent architecture composed of a planner, an evolver, and an evaluator. By combining operator abstraction with evolutionary search strategies, the framework enables explicit control over the trade-off between area and delay. Evaluated on the EPFL benchmark suite, the proposed method reduces area by 10.04% on average compared to ABC and outperforms mockturtle by 7.93%, achieving overall performance improvements ranging from 46.6% to 96.0%.

Abstract

Technology mapping is a critical yet challenging stage in logic synthesis. While Large Language Models (LLMs) have been applied to generate optimization scripts, their potential for core algorithm enhancement remains untapped. We introduce MappingEvolve, an open-source framework that pioneers the use of LLMs to directly evolve technology mapping code. Our method abstracts the mapping process into distinct optimization operators and employs a hierarchical agent-based architecture, comprising a Planner, Evolver, and Evaluator, to guide the evolutionary search. This structured approach enables strategic and effective code modifications. Experiments show our method significantly outperforms direct evolution and strong baselines, achieving 10.04\% area reduction versus ABC and 7.93\% versus mockturtle, with 46.6\%--96.0\% $S_{overall}$ improvement on EPFL benchmarks, while explicitly navigating the area--delay trade-off. Our code and data are available at https://github.com/Flians/MappingEvolve.

Problem

• technology mapping
• logic synthesis
• Large Language Models
• code evolution
• optimization

Innovation

• Technology Mapping
• Large Language Models
• Code Evolution
• Logic Synthesis
• Hierarchical Agent Architecture