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:20260422T143138Z
LOCATION:B304
DTSTART;TZID=America/New_York:20241117T170000
DTEND;TZID=America/New_York:20241117T171500
UID:submissions.supercomputing.org_SC24_sess745_ws_scalah104@linklings.com
SUMMARY:Sequences of Distributed Matrix-Vector Product for Very Large and 
 Very Sparse Irregular Matrices
DESCRIPTION:Maxence Vandromme, Nicolas Hochart, and Serge G. Petiton (Univ
 ersity of Lille, France); Jérôme Gurhem (Maison de la Simulation); and Miw
 ako Tsuji and Mitsuhisa Sato (RIKEN)\n\nWe study the performance behavior 
 of the sparse matrix-vector product operation in distributed computing env
 ironments, in the case of very large non-diagonal matrices where the nonze
 ro elements are placed irregularly across the matrix. In particular, we fo
 cus on the distributed storage of the result vector in cases where it beco
 mes too large to be stored fully on each process, and its redistribution b
 etween the iterations of a sequence of SpMV operations. We propose two met
 hods to this effect; one aims at minimizing the memory requirements of sto
 ring the result vector, the other optimizes the communications required fo
 r the redistribution. We perform large-scale experiments on the Fugaku sup
 ercomputer, in order to show the importance of taking into account the net
 work topology to correctly identify and address communication bottlenecks.
  The results show that the most efficient proposed method has a runtime se
 veral times faster than a non-optimal one on this topology.\n\nTag: Algori
 thms, Heterogeneous Computing\n\nRegistration Category: Workshop Reg Pass\
 n\nSession Chairs: Vassil Alexandrov (Hartree Centre, STFC); Jack Dongarra
  (University of Tennessee, Knoxville; Oak Ridge National Laboratory (ORNL)
 ); Erik Draeger (Lawrence Livermore National Laboratory (LLNL), Center for
  Applied Scientific Computing); Christian Engelmann (Oak Ridge National La
 boratory (ORNL)); and Dieter A. Kranzlmueller (Ludwig-Maxmilians-Universit
 ät München, Leibniz Supercomputing Centre (LRZ))\n\n
END:VEVENT
END:VCALENDAR