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:20210402T160103Z
LOCATION:Track 3
DTSTART;TZID=America/New_York:20201119T103000
DTEND;TZID=America/New_York:20201119T110000
UID:submissions.supercomputing.org_SC20_sess156_pap226@linklings.com
SUMMARY:Speeding Up SpMV for Power-Law Graph Analytics by Enhancing Locali
 ty & Vectorization
DESCRIPTION:Paper\n\nSpeeding Up SpMV for Power-Law Graph Analytics by Enh
 ancing Locality & Vectorization\n\nYesil, Heidarshenas, Morrison, Torrella
 s\n\nGraph analytics often target large-scale web and social networks, whi
 ch are typically power-law graphs. Graph algorithms can often be recast as
  generalized SpMV operations, making SpMV important to optimize for graph 
 analytics. However, executing SpMV on large-scale power-law graphs results
  in highly irregular memory access patterns with poor cache utilization. W
 orse, we find that existing locality and vectorization optimizations are l
 argely ineffective on modern out-of-order processors. To improve performan
 ce for power-law graphs on modern OOO processors, we propose Locality-Awar
 e Vectorization (LAV). LAV is a new approach that leverages the input’s po
 wer-law nature to extract locality and enable effective vectorization for 
 SpMV-like memory access patterns. \n\nWe evaluate LAV with several graphs
  on an Intel Skylake-SP processor, and find that LAV is faster than CSR (a
 nd prior approaches) by an average of 1.5x. LAV reduces the number of DRAM
  accesses by 35% on average with only 3.3% memory overhead.\n\nTag: Algori
 thms, Graph Algorithms, Linear Algebra\n\nRegistration Category: Tech Prog
 ram Reg Pass
END:VEVENT
END:VCALENDAR