BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20200129T163556Z
LOCATION:401-402-403-404
DTSTART;TZID=America/Denver:20191122T092000
DTEND;TZID=America/Denver:20191122T093500
UID:submissions.supercomputing.org_SC19_sess132_ws_p3hpc115@linklings.com
SUMMARY:An Approach for Indirectly Adopting a Performance Portability Laye
 r in Large Legacy Codes
DESCRIPTION:Workshop\n\nAn Approach for Indirectly Adopting a Performance 
 Portability Layer in Large Legacy Codes\n\nHolmen, Peterson, Berzins\n\nDi
 versity among supported architectures in current and emerging high perform
 ance computing systems, including those for exascale, makes portable codeb
 ases desirable. Portability of a codebase can be improved using a performa
 nce portability layer to provide access to multiple underlying programming
  models through a single interface. Direct adoption of a performance porta
 bility layer, however, poses challenges for large pre-existing software fr
 ameworks that may need to preserve legacy code and/or adopt other programm
 ing models in the future. This paper describes an approach for indirect ad
 option that introduces a framework-specific portability layer between the 
 application developer and the adopted performance portability layer to hel
 p improve legacy code support and long-term portability for future archite
 ctures and programming models. This intermediate layer uses loop-level, ap
 plication-level, and build-level components to ease adoption of a performa
 nce portability layer in large legacy codebases. Results are shown for two
  challenging case studies using this approach to make portable use of Open
 MP and CUDA via Kokkos in an asynchronous many-task runtime system, Uintah
 . These results show performance improvements up to 2.7x when refactoring 
 for portability and 2.6x when more efficiently using a node. Good strong-s
 caling to 442,368 threads across 1,728 Knights Landing processors are also
  shown using MPI+Kokkos at scale.\n\nTag: Workshop Reg Pass, Parallel Prog
 ramming Languages, Libraries, and Models, Performance, Portability, Produc
 tivity\n\nRegistration Category: Workshop Reg Pass, Parallel Programming L
 anguages, Libraries, and Models, Performance, Portability, Productivity
URL:https://sc19.supercomputing.org/presentation/?id=ws_p3hpc115&sess=sess
 132
END:VEVENT
END:VCALENDAR