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:20260422T000711Z
LOCATION:401-402
DTSTART;TZID=America/Denver:20231115T110000
DTEND;TZID=America/Denver:20231115T113000
UID:submissions.supercomputing.org_SC23_sess157_pap479@linklings.com
SUMMARY:Unified Communication Optimization Strategies for Sparse Triangula
 r Solver on CPU and GPU Clusters
DESCRIPTION:Yang Liu and Nan Ding (Lawrence Berkeley National Laboratory (
 LBNL)), Piyush Sao (Oak Ridge National Laboratory (ORNL)), and Samuel Will
 iams and Xiaoye Sherry Li (Lawrence Berkeley National Laboratory (LBNL))\n
 \nThis paper presents a unified framework for reducing communication costs
  of sparse triangular solvers (SpTRSV) on CPU and GPU clusters. The propos
 ed framework builds upon a 3D communication-avoiding process layout that d
 istributes a sparse triangular matrix into a 3D layout consisting of 2D gr
 ids. This work significantly reduces inter-process communication by replic
 ating computation and using sparse allreduce operations across the 2D grid
 s. This also allows for integration of a number of communication-optimized
  2D SpTRSV algorithms including binary communication tree-based CPU algori
 thms and one-sided GPU communication (e.g., NVSHMEM)-based algorithms. Wit
 h all these communication reduction schemes, the resulting SpTRSV exhibits
  significantly better scalability than existing works on leadership CPU an
 d CPU clusters such as Cori, Perlmutter and Crusher.\n\nTag: Accelerators,
  Algorithms, Linear Algebra\n\nRegistration Category: Tech Program Reg Pas
 s\n\nAward Finalist: Best Paper Finalist\n\nReproducibility Badges: Artifa
 ct Available, Artifact Functional, Results Reproduced\n\nSession Chair: Ha
 rtwig Anzt (Technical University of Munich; University of Tennessee, Knoxv
 ille)\n\n
END:VEVENT
END:VCALENDAR