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:20211207T055411Z
LOCATION:Online
DTSTART;TZID=America/Chicago:20211119T111000
DTEND;TZID=America/Chicago:20211119T113000
UID:submissions.supercomputing.org_SC21_sess329_ws_pawatm103@linklings.com
SUMMARY:Optimization of Asynchronous Communication Operations through Eage
 r Notifications
DESCRIPTION:Workshop\n\nOptimization of Asynchronous Communication Operati
 ons through Eager Notifications\n\nKamil, Bonachea\n\nUPC++ is a C++ libra
 ry implementing the Asynchronous Partitioned Global Address Space (APGAS) 
 model. We propose an enhancement to the completion mechanisms of UPC++ use
 d to synchronize communication operations that is designed to reduce overh
 ead for on-node operations. Our enhancement permits eager delivery of comp
 letion notification in cases where the data transfer semantics of an opera
 tion happen to complete synchronously, for example due to the use of share
 d-memory bypass. This semantic relaxation allows removing significant over
 head from the critical path of the implementation in such cases.\n\nWe eva
 luate our results on three different representative systems using a combin
 ation of microbenchmarks and five variations of the the HPCChallenge Rando
 mAccess benchmark implemented in UPC++ and run on a single node to accentu
 ate the impact of locality. We find that in RMA versions of the benchmark 
 written in a straightforward manner (without manually optimizing for local
 ity), the new eager notification mode can provide up to a 25% speedup when
  synchronizing with promises and up to a 13.5x speedup when synchronizing 
 with conjoined futures. We also evaluate our results using a graph-matchin
 g application written with UPC++ RMA communication, where we measure overa
 ll speedups of as much as 11% in single-node runs of the unmodified applic
 ation code, due to our transparent enhancements.\n\nTag: Online Only, Appl
 ications, Extreme Scale Computing, Heterogeneous Systems, Parallel Program
 ming Languages and Models, Software Engineering\n\nRegistration Category: 
 Workshop Reg Pass
END:VEVENT
END:VCALENDAR