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:20260422T000712Z
LOCATION:301-302-303
DTSTART;TZID=America/Denver:20231116T153000
DTEND;TZID=America/Denver:20231116T160000
UID:submissions.supercomputing.org_SC23_sess162_pap304@linklings.com
SUMMARY:Enhance the Strong Scaling of LAMMPS on Fugaku
DESCRIPTION:Jianxiong Li, Tong Zhao, Zhuoqiang Guo, and Shunchen Shi (Stat
 e Key Lab of Processors, Institute of Computing Technology, Chinese Academ
 y of Sciences; University of Chinese Academy of Sciences); Lijun Liu (Depa
 rtment of Mechanical Engineering, Graduate School of Engineering, Osaka Un
 iversity.); and Guangming Tan, Weile Jia, Guojun Yuan, and Zhan Wang (Stat
 e Key Lab of Processors, Institute of Computing Technology, Chinese Academ
 y of Sciences; University of Chinese Academy of Sciences)\n\nPhysical phen
 omenon such as protein folding requires simulation up to microseconds of p
 hysical time, which directly corresponds to the strong scaling of molecula
 r dynamics(MD) on modern supercomputers. In this paper, we present a highl
 y scalable implementation of the state-of-the-art MD code LAMMPS on Fugaku
  by exploiting the 6D mesh/torus topology of the TofuD network. Based on o
 ur detailed analysis of the MD communication pattern, we first adapt coars
 e-grained peer-to-peer ghost-region communication with uTofu interface, th
 en further improve the scalability via fine-grained thread pool. Finally, 
 Remote direct memory access (RDMA) primitives are utilized to avoid buffer
  overhead. Numerical results show that our optimized code can reduce 77% o
 f the communication time, improving the performance of baseline LAMMPS by 
 a factor of 2.9x and 2.2x for Lennard-Jones and embedded-atom method poten
 tials when scaling to 36, 846 computing nodes. Our optimization techniques
  can also benefit other applications with stencil or domain decomposition 
 methods.\n\nTag: Accelerators, Applications, Architecture and Networks, Mo
 deling and Simulation\n\nRegistration Category: Tech Program Reg Pass\n\nR
 eproducibility Badges: Artifact Available\n\nSession Chair: Sudip Seal (Oa
 k Ridge National Laboratory (ORNL))\n\n
END:VEVENT
END:VCALENDAR