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:20220812T074335Z
LOCATION:Singapore Room
DTSTART;TZID=Europe/Stockholm:20220629T113000
DTEND;TZID=Europe/Stockholm:20220629T120000
UID:submissions.pasc-conference.org_PASC22_sess156_msa227@linklings.com
SUMMARY:FourCastNet - An Example DL Architecture for Scaling Deep Learning
  Applications to Large HPC Systems
DESCRIPTION:Minisymposium\n\nFourCastNet - An Example DL Architecture for 
 Scaling Deep Learning Applications to Large HPC Systems\n\nKurth, Messmer\
 n\nWe are going to present the challenges of scaling deep learning applica
 tions to large HPC systems on the example of FourCastNet.FourCastNet is a 
 Fourier Neural Operator (FNO) based neural network which employs spectral 
 convolutions to generate competitive weather predictions for up to 8 days 
 into the future. In addition to classical parallelization techniques commo
 nly used in deep learning applications, the specific architectural structu
 re of FourCastNet allows for various parallelization techniques which are 
 more commonly found in traditional HPC simulations. <br />We will discuss 
 the advantages and disadvantages of the various alternatives and predict s
 calability on the latest generation of NVIDIA GPUs using a performance mod
 el. We will further discuss the prospects of utilizing GPU architectural a
 dvancements for traditional HPC applications which employ spectral solvers
 .\n\nDomain: Computer Science and Applied Mathematics
END:VEVENT
END:VCALENDAR