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:20181221T160728Z
LOCATION:D172
DTSTART;TZID=America/Chicago:20181112T155500
DTEND;TZID=America/Chicago:20181112T162000
UID:submissions.supercomputing.org_SC18_sess168_ws_ia112@linklings.com
SUMMARY:A Block-Oriented, Parallel, and Collective Approach to Sparse Inde
 finite Preconditioning on GPUs
DESCRIPTION:Workshop\nArchitectures, Data Analytics, Graph Algorithms, Wor
 kshop Reg Pass\n\nA Block-Oriented, Parallel, and Collective Approach to S
 parse Indefinite Preconditioning on GPUs\n\nThuerck, Naumov, Goesele, Garl
 and\n\nLarge sparse symmetric indefinite matrices are notoriously hard to 
 precondition. They often lack  diagonal dominance and exhibit Schur-comple
 ments that render zero fill-in factorization preconditioning ineffective. 
 Pivoting, a necessity for stable LDLt factorizations, complicates parallel
  approaches that can take advantage of the latest massively-parallel HPC h
 ardware such as GPUs. We present an approach based on ad-hoc blocking and 
 reordering strategies that allows local, independent collective-oriented p
 rocessing of small dense blocks. A hybrid block-memory layout compensates 
 for irregular memory access patterns found in sparse matrices. Our method 
 allows restricted fill-in, supernodal pivoting and a dual threshold droppi
 ng strategy at little additional cost. It delivers robust preconditioners 
 that in our experiments obtain an average speedup of ~6x even for tough ma
 trices from optimization problems.
URL:https://sc18.supercomputing.org/presentation/?id=ws_ia112&sess=sess168
END:VEVENT
END:VCALENDAR