BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260402T024533Z
LOCATION:3006\, Level 3
DTSTART;TZID=America/Los_Angeles:20250623T154500
DTEND;TZID=America/Los_Angeles:20250623T160000
UID:dac_DAC 2025_sess135_RESEARCH339@linklings.com
SUMMARY:IRGNN: A Graph-based Framework Integrating Numerical Solution and 
 Point Cloud for Static IR Drop Prediction
DESCRIPTION:Feng Guo, Yueyue Xi, Jianwang Zhai, Jingyu Jia, Jiawei Liu, Ka
 ng Zhao, and Chuan Shi (Beijing University of Posts and Telecommunications
 )\n\nWith the continued scaling of integrated circuits (ICs), IR drop anal
 ysis for on-chip power grids (PGs) is crucial but increasingly computation
 ally demanding. \nTraditional numerical methods deliver high accuracy but 
 are prohibitively time-intensive, while various machine learning (ML) meth
 ods have been introduced to alleviate these computational burdens. \nHowev
 er, most CNN-based methods ignore the fine structure and topological infor
 mation of PGs, and face interpretability or scalability issues.\nIn this w
 ork, we propose a novel graph-based framework, IRGNN, leveraging the PG to
 pology with the integration of numerical solutions and point clouds. \nOur
  framework applies a numerical solver, AMG-PCG, to generate rough numerica
 l solutions as a reliable interpretability foundation for ML.\nThen, to ca
 pture PG topology, we regard nodes of PG as point clouds and extract point
  cloud features, and we introduce a novel graph structure, IRGraph.\nFurth
 ermore, a novel graph-based model IRGNN is designed, incorporating a desig
 ned neighbor distance attention (NDA) layer for distance-aware PG features
  aggregation and graph transformer (GT) layer to capture global informatio
 n.\nIt should be noted that our framework can analyze the IR drop of each 
 node in PG, which CNN-based methods cannot do.\nExperimental evaluations d
 emonstrate that our framework achieves significantly higher accuracy than 
 previous CNN-based approaches and numerical solvers while substantially re
 ducing computation time.\n\nTopics: EDA\n\nTracks: EDA4: Power Analysis an
 d Optimization\n\nSession Chairs: Prabal Basu (Cadence Design Systems, Inc
 .) and Noel Daniel Gundi (Utah State University)\n\n
END:VEVENT
END:VCALENDAR