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:20260522T150128Z
LOCATION:D175
DTSTART;TZID=America/Chicago:20181111T093600
DTEND;TZID=America/Chicago:20181111T100000
UID:submissions.supercomputing.org_SC18_sess155_ws_waccpd102@linklings.com
SUMMARY:Heterogeneous Programming and Optimization of Gyrokinetic Toroidal
  Code Using Directives
DESCRIPTION:Wenlu Zhang (Institute of Physics, Chinese Academy of Sciences
 ; University of California, Irvine)\n\nThe latest production version of th
 e fusion particle simulation code, Gyrokinetic Toroidal Code (GTC), has be
 en ported to and optimized for the next generation exascale GPU supercompu
 ting platform. Heterogeneous programming using directives has been utilize
 d to fuse and thus balance the continuously implemented physical capabilit
 ies and rapidly evolving software/hardware systems. The original code has 
 been refactored to a set of unified functions/calls to enable the accelera
 tion for all the species of particles. Binning and GPU texture caching tec
 hnique have also been used to boost the performance of the particle push a
 nd shift operations. In order to identify the hotspots, the GPU version of
  the GTC code was the first benchmarked on up to 8000 nodes of the Titan s
 upercomputer, which shows about 2–3 times overall speedup comparing NVidia
  M2050 GPUs to Intel Xeon X5670 CPUs. This Phase I optimization was follow
 ed by further optimizations in Phase II, where single-node tests show an o
 verall speedup of about 34 times on SummitDev and 7.9 times on Titan. The 
 real physics tests on Summit machine showed impressive scaling properties 
 that reaches roughly 50% efficiency on 928 nodes of Summit. The GPU+CPU sp
 eed up from purely CPU is over 20 times, leading to an unparalleled speed.
 \n\nTag: Accelerators, Heterogeneous Systems, Parallel Programming Languag
 es, Libraries, and Models\n\nRegistration Category: Workshop Reg Pass\n\nS
 ession Chairs: Sunita Chandrasekaran (University of Delaware); Guido Jucke
 land (Helmholtz-Zentrum Dresden-Rossendorf (HZDR), TUD Dresden University 
 of Technology); and Sandra Wienke (RWTH Aachen University)\n\n
END:VEVENT
END:VCALENDAR