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:20260422T143140Z
LOCATION:B304
DTSTART;TZID=America/New_York:20241117T163000
DTEND;TZID=America/New_York:20241117T164500
UID:submissions.supercomputing.org_SC24_sess745_ws_scalah108@linklings.com
SUMMARY:A Performance Portable Multi-GPU Implementation of 3D Euler Equati
 ons using ProtoX and IRIS
DESCRIPTION:Het Mankad and Mohammad Alaul Haque Monil (Oak Ridge National 
 Laboratory (ORNL)), Sanil Rao (Carnegie Mellon University), Phillip Colell
 a and Brian Van Straalen (Lawrence Berkeley National Laboratory (LBNL)), F
 ranz Franchetti (Carnegie Mellon University), and Jeffrey Vetter (Oak Ridg
 e National Laboratory (ORNL))\n\nDomain scientists in the field of computa
 tional science often face challenges in developing optimized code for high
 -performance computing, especially GPUs. Considering the increase of heter
 ogeneity in a node of HPC computing facilities, there is a demand to devel
 op performance portable solutions for the core computation kernels in a sc
 ientific library. We demonstrate a\nperformance portable multi-GPU solutio
 n for an implementation of the Euler equations using ProtoX and IRIS. Prot
 oX is a domain-specific language using a partial differential equation lib
 rary called Proto as its front end and SPIRAL, a code generation system, i
 n its back end to generate optimized kernels for different architectures. 
 These kernels are orchestrated through the intelligent runtime system --- 
 IRIS to provide portability. Two levels of optimizations within IRIS, name
 ly DAG and task fusion, are explored to efficiently utilize computing reso
 urces in a multi-GPU environment. Performance improvement through these op
 timizations is showcased on AMD and NVIDIA GPUs.\n\nTag: Algorithms, Heter
 ogeneous Computing\n\nRegistration Category: Workshop Reg Pass\n\nSession 
 Chairs: Vassil Alexandrov (Hartree Centre, STFC); Jack Dongarra (Universit
 y of Tennessee, Knoxville; Oak Ridge National Laboratory (ORNL)); Erik Dra
 eger (Lawrence Livermore National Laboratory (LLNL), Center for Applied Sc
 ientific Computing); Christian Engelmann (Oak Ridge National Laboratory (O
 RNL)); and Dieter A. Kranzlmueller (Ludwig-Maxmilians-Universität München,
  Leibniz Supercomputing Centre (LRZ))\n\n
END:VEVENT
END:VCALENDAR