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:20181221T160906Z
LOCATION:C145
DTSTART;TZID=America/Chicago:20181114T150000
DTEND;TZID=America/Chicago:20181114T170000
UID:submissions.supercomputing.org_SC18_sess468_spost132@linklings.com
SUMMARY:Modeling Single-Source Shortest Path Algorithm Dynamics to Control
  Performance and Power Tradeoffs
DESCRIPTION:ACM Student Research Competition, Poster\nStudent Program, Tec
 h Program Reg Pass, ACM Student Research Competition\n\nModeling Single-So
 urce Shortest Path Algorithm Dynamics to Control Performance and Power Tra
 deoffs\n\nKaramati, Young, Vuduc\n\nThis work presents a new methodology t
 o improve the performance of parallel algorithms by tuning the amount of a
 vailable parallelism for execution throughout the runtime. As such, we exp
 ose key parameters controlling the performance and parallelism of the algo
 rithm and build a software-based controller with the objective of maintain
 ing the optimal performance. Our controller allows for tuning the level of
  parallelism executed in each time epoch to optimize for performance while
  preserving power usage. More specifically, our experimental evaluation fo
 cuses on a tunable variation of a GPU-based delta-stepping algorithm for c
 omputing the single-source shortest path (SSSP); As the available parallel
 ism for the delta-stepping SSSP is highly irregular and strongly input-dep
 endent, our extensive experiments show that average power can be reduced w
 hile average parallelism is increased. This increase in average parallelis
 m provides substantial energy savings, independent of the hardware.
URL:https://sc18.supercomputing.org/presentation/?id=spost132&sess=sess468
END:VEVENT
END:VCALENDAR

