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:20241120T082409Z
LOCATION:HG E 1.2
DTSTART;TZID=Europe/Stockholm:20240604T173000
DTEND;TZID=Europe/Stockholm:20240604T180000
UID:submissions.pasc-conference.org_PASC24_sess114_msa266@linklings.com
SUMMARY:Optimizing CRK-HACC for Performance Portability Using SYCL
DESCRIPTION:Minisymposium\n\nEsteban Rangel (Argonne National Laboratory)\
 n\nIn this talk, we discuss the development of the SYCL implementation of 
 CRK-HACC, an extreme-scale cosmological simulation code with physics for r
 esolving gas hydrodynamics. We describe our CUDA-to-SYCL migration pipelin
 e for producing function objects and detail how we achieved a high level o
 f “performance portability” across GPUs from AMD, Intel, and NVIDIA, requi
 ring us to develop an abstraction for multiple “shuffle” operations: the s
 ycl::select_from_group function from SYCL 2020, a shuffle operation emulat
 ed via work-group local memory, and a highly specialized shuffle operation
  implemented for Intel GPUs in assembly (vISA). To facilitate code maintai
 nability we also created abstractions for host-side code that is shared ac
 ross HIP, SYCL, and CUDA. We believe our techniques will generalize well t
 o other application domains and provide a balance of maintainability and p
 erformance portability.\n\nDomain: Engineering, Computational Methods and 
 Applied Mathematics\n\nSession Chair: Szilárd Páll (KTH Royal Institute of
  Technology, PDC Center for High Performance Computing)
END:VEVENT
END:VCALENDAR