Data and code: A Zero-Fitted-Parameter Physics-Based Estimator for Early Integrated Voltage Regulator Placement Ranking

A physics-based early estimator ranks candidate placements for on-chip integrated voltage regulators (IVRs) using a resistive-mesh (Modified Nodal Analysis) model built from real, extracted power-delivery-network geometry and GF180MCU process-design-kit constants, with zero parameters fit to any measurement. The estimator is validated end-to-end against OpenROAD's PDNSim IR-drop analysis engine across three structurally distinct real RTL designs (PicoRV32, AES, SHA-256), each evaluated over 120 candidate placements at two regulator counts. This repository contains the raw per-candidate experimental data (120 candidates per benchmark per configuration), the real extracted PDN geometry and instance-area data each benchmark's estimator run depends on, the frozen estimator implementation (MD5-verified, unmodified across all three benchmarks), and the analysis scripts that reproduce every quantitative result reported in the accompanying manuscript. See README.md in the archive for full column documentation and reproduction instructions.