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:20210916T132447Z
LOCATION:Louis Favre
DTSTART;TZID=Europe/Stockholm:20210705T153000
DTEND;TZID=Europe/Stockholm:20210705T160000
UID:submissions.pasc-conference.org_PASC21_sess111_msa333@linklings.com
SUMMARY:Solving Nonlinear Partial Differential Equations on GPU Supercompu
 ters Using Julia
DESCRIPTION:Minisymposium\n\nSolving Nonlinear Partial Differential Equati
 ons on GPU Supercomputers Using Julia\n\nOmlin, Räss, Keepfer, Kwasniewski
 , Malvoisin...\n\nWe present a self-contained approach for the development
  of massively scalable multi-GPU solvers for coupled nonlinear systems of 
 partial differential equations (PDE) in Julia. The approach encompasses nu
 merics, implementation and performance evaluation. It relies on the usage 
 of a powerful stencil-based iterative method which enables to efficiently 
 converge to the time-dependent implicit solution for strongly nonlinear pr
 oblems; the method optimally suits both shared and distributed memory para
 llelism. The implementation approach enables a straightforward development
  of a single Julia code that can be readily deployed on a single CPU threa
 d or on thousands of GPUs/CPUs. The performance evaluation is conducted wi
 th a simple memory throughput-based metric for iterative PDE solvers. We d
 emonstrate the wide applicability of our approach by showcasing several 2-
 D and 3-D Multi-GPU PDE solvers as, e.g., a solver for spontaneous nonline
 ar multi-physics porous flow localization in 3-D. As reference, the latter
  solver was ported from MPI+CUDA C to Julia and achieves 95% of the perfor
 mance of the original solver and a nearly ideal parallel efficiency on tho
 usands of NVIDIA Tesla P100 GPUs on the hybrid Cray XC-50 "Piz Daint" supe
 rcomputer at the Swiss National Supercomputing Centre, CSCS.\n\nDomain: CS
  and Math, Climate and Weather, Physics, Solid Earth Dynamics
END:VEVENT
END:VCALENDAR