BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20241120T082409Z
LOCATION:HG E 3
DTSTART;TZID=Europe/Stockholm:20240604T160000
DTEND;TZID=Europe/Stockholm:20240604T163000
UID:submissions.pasc-conference.org_PASC24_sess139_msa196@linklings.com
SUMMARY:SimAI-Bench: A Performance Benchmarking Tool for Coupled Simulatio
 n and AI Workflows
DESCRIPTION:Minisymposium\n\nRiccardo Balin, Shivam Barwey, Ramesh Balakri
 shnan, Bethany Lusch, Saumil Patel, Tom Uram, and Venkatram Vishwanath (Ar
 gonne National Laboratory)\n\nIn situ AI/ML workflows, in which ML tasks a
 re coupled to an ongoing simulation, are an attractive new paradigm for de
 veloping robust and predictive surrogate models for accelerating time to s
 cience by steering simulation ensembles and replacing expensive computatio
 ns. In the world of high performance computing (HPC), these workflows requ
 ire scalable and efficient solutions to integrate the rapidly evolving eco
 system of ML frameworks with traditional simulation codes by transferring 
 large volumes of data between the various components. To address these iss
 ues, several libraries have recently emerged from groups in industry, acad
 emia, and national labs. In this talk, we introduce SimAI-Bench – a new to
 ol for benchmarking and comparing the performance of different coupled sim
 ulation and AI/ML workflows on current and future HPC systems. In particul
 ar, the talk will focus on workflows for in situ training of graph neural 
 network (GNN) surrogate models from ongoing computational fluid dynamic (C
 FD) simulations, requiring the transfer of training data between the two c
 omponents. We will discuss how different open-source libraries enable such
  workflows and compare their data transfer performance and scaling efficie
 ncy on the Aurora supercomputer at the Argonne Leadership Computing Facili
 ty.\n\nDomain: Chemistry and Materials, Climate, Weather, and Earth Scienc
 es, Engineering, Computational Methods and Applied Mathematics\n\nSession 
 Chair: Alessandro Rigazzi (HPE)
END:VEVENT
END:VCALENDAR