BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260522T150116Z
LOCATION:D171/173
DTSTART;TZID=America/Chicago:20181116T103000
DTEND;TZID=America/Chicago:20181116T104500
UID:submissions.supercomputing.org_SC18_sess145_ws_p3hpc104@linklings.com
SUMMARY:Performance Portability of an Unstructured Hydrodynamics Mini-Appl
 ication
DESCRIPTION:Timothy R. Law (University of Warwick); Robert Kevis and Seimo
 n Powell (Atomic Weapons Establishment (AWE), UK); James Dickson (Universi
 ty of Warwick); Satheesh Maheswaran and J. A. Herdman (Atomic Weapons Esta
 blishment (AWE), UK); and Stephen A. Jarvis (University of Warwick)\n\nIn 
 this work we study the parallel performance portability of BookLeaf: a rec
 ent 2D unstructured hydrodynamics mini-application. The aim of BookLeaf is
  to provide a self-contained and representative testbed for exploration of
  the modern hydrodynamics application design-space.\n\nWe present a previo
 usly unpublished reference C++11 implementation of BookLeaf parallelised w
 ith MPI, alongside hybrid MPI+OpenMP and MPI+CUDA versions, and two implem
 entations using C++11 performance portability frameworks: Kokkos and RAJA,
  which both target a variety of parallel back-ends. We assess the scalabil
 ity of our implementations on the ARCHER Cray XC30 up to 4096 nodes (98,30
 4 cores) and on the Ray EA system at Lawrence Livermore National Laborator
 y up to 16 nodes (64 Tesla P100 GPUs), with a particular focus on the over
 heads introduced by Kokkos and RAJA relative to our handwritten OpenMP and
  CUDA implementations. We quantify the performance portability achieved by
  our Kokkos and RAJA implementations across five modern architectures usin
 g a metric previously introduced by Pennycook et al.\n\nWe find that our B
 ookLeaf implementations all scale well, in particular the hybrid configura
 tions (the MPI+OpenMP variant achieves a parallel efficiency above 0.8 run
 ning on 49,152 cores). The Kokkos and RAJA variants exhibit competitive pe
 rformance in all experiments, however their CPU performance is best in mem
 ory-bound situations where the overhead introduced by the frameworks is pa
 rtially shadowed by the need to wait for data. The overheads seen in the G
 PU experiments are extremely low. We observe overall performance portabili
 ty scores of 0.928 for Kokkos and 0.876 for RAJA.\n\nTag: Heterogeneous Sy
 stems, Performance\n\nRegistration Category: Workshop Reg Pass\n\nSession 
 Chair: Rob Neely (Lawrence Livermore National Laboratory)\n\n
END:VEVENT
END:VCALENDAR