Language models improve analog circuit layout repairs with cost-aware planning
Who Plans, Who Edits? Verified, Cost-Aware Division of Labor for LLM-Driven Analog Layout Repair
Hardware ArchitectureArtificial Intelligence
Summary
Fixing complex analog circuit designs by hand is slow and expensive. The authors show that using different language models to plan and carry out repairs, combined with checks after each step, can fix these designs more efficiently. They also teach a controller to pick which model to use and when to escalate to stronger models. This approach speeds up repairs while keeping costs low and produces fixes as accurate as those done by humans.
What this means in practice
- •For analog circuit designers: Automate layout repair processes by using a model-based controller to reduce design time and costs while maintaining human-level repair quality.
- •For hardware testing teams: Use the verification-driven model workflow to catch layout errors automatically after each edit, ensuring only valid designs proceed to manufacturing.
- •For chip manufacturing automation engineers: Integrate language-model driven repair workflows to streamline post-layout verification and correction steps in semiconductor fabrication flows.$Commercial implications: Enables commercial EDA tools offering automated verified layout repair, reducing time and costs for chip makers.
Authors
Xilai Dai, Saeed Mohammadi
Abstract
Language-model agents can repair analog layouts when sign-off tools check every edit, but which model should plan a repair, which should edit it, and at what cost remain open questions. In our tool-grounded repair agent, models only choose bounded layout operators; sign-off tools rerun each task's contract checks after every edit, so failing layouts are never accepted. On layout-versus-schematic (LVS), design-rule checking (DRC), and post-layout repairs of three designs (a production high-voltage digital-to-analog converter (DAC), a 5-V oscillator, and a SKY130 operational transconductance amplifier (OTA)) with six tier-matched models from two vendors, 182 blind runs show that cheap models execute plans but rarely succeed alone (the cheapest model: 1 of 11 runs alone, 16 of 17 under a flagship planner; p < 0.001), cheap planners fail more often, and no fixed assignment is cheapest on every task. We propose a learning controller that selects models for each role and the escalation order from the first tool evaluation and updates Bayesian estimates after every attempt. With rules and initial estimates frozen before testing, the controller fixed 7 of 8 episodes on four sealed held-out tasks, matching the OpenAI flagship alone, at 15-40% of the flagship's spend per fix on three tasks but 8% higher spend overall. The cheapest mid-tier model alone fixed 6 of 8 at the lowest overall spend per fix but missed the hardest task. All 32 extracted DAC repairs match the human-guided LVS repair's integral nonlinearity (INL) within 0.005 least significant bits (LSB) at every code.