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:20260522T162632Z
LOCATION:Bldg. 6 - Room 002
DTSTART;TZID=Europe/Stockholm:20260629T113000
DTEND;TZID=Europe/Stockholm:20260629T120000
UID:submissions.pasc-conference.org_PASC26_sess166_pap115@linklings.com
SUMMARY:Kokkos Comm: Performance Portable Communication for Distributed Ko
 kkos Applications
DESCRIPTION:C. Nicole Avans (Sandia National Laboratories, Tennessee Techn
 ological University); Gabriel Dos Santos, Cédric Chevalier, Hugo Taboada, 
 and Marc Pérache (CEA, Université Paris-Saclay); Matthew G. F. Dosanjh, St
 ephen L. Olivier, and Carl Pearson (Sandia National Laboratories); Evan D.
  Suggs (Tennessee Technological University); Vivek Kale (Sandia National L
 aboratories); and Anthony Skjellum (Tennessee Technological University)\n\
 nThis paper introduces Kokkos Comm, a new library and API specification fo
 r enhanced performance, portability, and productivity. Kokkos is a widely 
 used C++ performance portability ecosystem that addresses performance port
 ability on-node through advanced C++ metaprogramming, including GPU and Op
 enMP targeting support. Kokkos Comm addresses challenges of integrating Ko
 kkos with distributed memory programming models. Kokkos Comm alleviates ac
 cidental complexity associated with coordinating non-blocking communicatio
 n operations and \Kokkos{} execution spaces, as well as handling non-conti
 guous Kokkos::Views. Automation of common low-level, error-prone implement
 ation details, such as packing and unpacking of non-contiguous data, incre
 ases programmer productivity while decreasing code complexity with the add
 ed potential for higher performance (e.g., vs. MPI derived datatypes). Fur
 ther, this library serves as a platform for researching improved methods f
 or managing non-contiguous data and exploring new communication APIs with 
 performance portability across various underlying transports and accelerat
 ors. For instance, Kokkos Comm enables use of varied internal implementati
 ons of the data transfer functionality (e.g., MPI RMA, NCCL, SHMEM dialect
 s, libfabric), while maintaining overall support for MPI elsewhere in an a
 pplication (e.g., ScaLAPACK).\n\nSession Chair: Sriramkrishnan Muralikrish
 nan (Forschungszentrum Jülich)\n\n
END:VEVENT
END:VCALENDAR