BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260422T000713Z
LOCATION:403-404
DTSTART;TZID=America/Denver:20231114T110000
DTEND;TZID=America/Denver:20231114T113000
UID:submissions.supercomputing.org_SC23_sess177_pap408@linklings.com
SUMMARY:Large-Scale Simulation of Structural Dynamics Computing on GPU Clu
 sters
DESCRIPTION:Yumeng Shi (Computer Network Information Center, Chinese Acade
 my of Sciences); Ningming Nie (Computer Network Information Center, Chines
 e Academy of Sciences; University of Chinese Academy of Sciences); Shunde 
 Li and Jue Wang (Computer Network Information Center, Chinese Academy of S
 ciences); Kehao Lin (Hangzhou Dianzi University; Computer Network Informat
 ion Center, Chinese Academy of Sciences); Chunbao Zhou (Computer Network I
 nformation Center, Chinese Academy of Sciences); Shigang Li (School of Com
 puter Science, Beijing University of Posts and Telecommunications); Kehan 
 Yao (Hangzhou Dianzi University); Yangde Feng (Computer Network Informatio
 n Center, Chinese Academy of Sciences); Yan Zeng (Hangzhou Dianzi Universi
 ty); Fang Liu and Yangang Wang (Computer Network Information Center, Chine
 se Academy of Sciences); and Yue Gao (China Institute of Atomic Energy)\n\
 nStructural dynamics simulation plays an important role in research on rea
 ctor design and complex engineering. The Hybrid Total Finite Element Teari
 ng and Interconnecting (HTFETI) method combined with Newmark method is an 
 efficient way to solve large-scale structural dynamics problems. However, 
 the sparse direct solver and the load imbalance caused by inconsistent den
 sity models are two critical issues limiting the performance and the scala
 bility of structural dynamics computing. For the former, we propose an eff
 icient variable-size batched method to accelerate SpMV on GPUs. For the la
 tter, we establish an online performance prediction model, based on which 
 we then design a novel inter-cluster subdomain fine-tuning algorithm to ba
 lance the workload of HTFETI parallel computing. We are the first to achie
 ve the high-fidelity structural dynamics simulation of China Experimental 
 Fast Reactor core assembly with up to 53.4 billion grids. The weak and str
 ong scalability efficiencies reach 91.77% and 86.13% on 12,800 GPUs, respe
 ctively.\n\nTag: Accelerators, Applications, Modeling and Simulation\n\nRe
 gistration Category: Tech Program Reg Pass\n\nAward Finalist: Best Paper F
 inalist, Best Student Paper Finalist\n\nReproducibility Badges: Artifact A
 vailable, Artifact Functional, Results Reproduced\n\nSession Chair: Sameh 
 Abdulah (King Abdullah University of Science and Technology (KAUST))\n\n
END:VEVENT
END:VCALENDAR