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:20260924T172455Z
LOCATION:B309
DTSTART;TZID=America/New_York:20241119T163000
DTEND;TZID=America/New_York:20241119T170000
UID:submissions.supercomputing.org_SC24_sess401_pap649@linklings.com
SUMMARY:HiRace: Accurate and Fast Data-Race Checking for GPU Programs
DESCRIPTION:John Jacobson (University of Utah), Martin Burtscher (Texas St
 ate University), and Ganesh Gopalakrishnan (University of Utah)\n\nData ra
 ces are egregious concurrency bugs that are especially problematic in perf
 ormance-oriented GPU codes where large thread counts and multiple shared m
 emory regions tend to exacerbate them. In this work, we present a new dyna
 mic data-race checker called HiRace, whose key novelty is an innovative st
 ate machine designed to capitalize on the bulk-synchronous hierarchical GP
 U programming model. This state machine condenses an arbitrarily long acce
 ss history into a constant-size state. We evaluate HiRace on a large, cali
 brated data-race benchmark suite. In over 3,500 studied executions of 580 
 CUDA kernels, 346 of which contain data races, we found HiRace to detect r
 aces missed by other tools without raising false alarms and to be more tha
 n 10 times faster on average than the current state of the art with half t
 he memory overhead.\n\nTag: Accelerators, Compilers, Heterogeneous Computi
 ng, Performance Evaluation and/or Optimization Tools\n\nRegistration Categ
 ory: Tech Program Reg Pass\n\nSession Chair: Sascha Hunold (Technical Univ
 ersity of Vienna)\n\n
END:VEVENT
END:VCALENDAR