BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20250626T234542Z
LOCATION:B306
DTSTART;TZID=America/New_York:20241118T144600
DTEND;TZID=America/New_York:20241118T150000
UID:submissions.supercomputing.org_SC24_sess751_ws_p3hpc109@linklings.com
SUMMARY:Performance Portable Optimizations of an Ice-sheet Modeling Code o
 n GPU-supercomputers
DESCRIPTION:Oscar Antepara and Samuel Williams (Lawrence Berkeley National
  Laboratory (LBNL)) and Max Carlson and Jerry Watkins (Sandia National Lab
 oratories)\n\nIn this paper, we present GPU-optimizations for an ice-sheet
  modeling code known as MPAS-Albany Land Ice (MALI). MALI is a C++ templat
 e code that leverages Kokkos programming model for portability and Trilino
 s library for data structures, nonlinear and linear solvers. Performance o
 f the most expensive kernel is assessed via the Roofline model to highligh
 t the potential for code improvement according to the underlying GPU archi
 tecture. We perform optimizations consisting of loop fusion, loop optimiza
 tions and local accumulation to productively and portably attain an overal
 l speedup of 3$\times$ in either NVIDIA and AMD GPU. We analyze the perfor
 mance gains using a time-oriented performance portability model based on t
 ime per invocation and GPU data movement. Results show an increment betwee
 n 20\% and 50\% on the performance portability metric by improving data lo
 cality and highlights the importance of optimizing GPU-ported scientific a
 pplications to maximize memory bandwidth and minimize data movement on mod
 ern supercomputers.\n\nTag: Performance Optimization, Programming Framewor
 ks and System Software\n\nRegistration Category: Workshop Reg Pass\n\nSess
 ion Chairs: CJ Newburn (NVIDIA Corporation), Scott J. Parker (Argonne Nati
 onal Laboratory (ANL)), John Pennycook (Intel Corporation), and Kenneth We
 iss (Lawrence Livermore National Laboratory (LLNL))\n\n
END:VEVENT
END:VCALENDAR