BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20191206T142455Z
LOCATION:Substanz 1\, 2
DTSTART;TZID=Europe/Stockholm:20180627T083000
DTEND;TZID=Europe/Stockholm:20180627T090000
UID:isc_hpc_ISC High Performance 2018_sess161_pap115@linklings.com
SUMMARY:The Pitfalls of Provisioning Exascale Networks: A Trace Replay Ana
 lysis for Understanding Communication Performance
DESCRIPTION:Research Paper\n\nThe Pitfalls of Provisioning Exascale Networ
 ks: A Trace Replay Analysis for Understanding Communication Performance\n\
 nKenny, Knight, Sargsyan, Wilke, Michelogiannakis\n\nData movement is cons
 idered the main performance concern for exascale, including both on-node m
 emory and off-node network communication.  Indeed, many application traces
  show significant time spent in MPI calls, potentially indicating that fas
 ter networks must be provisioned for scalability.\nHowever, equating MPI t
 imes with network communication delays ignores synchronization delays and 
 software overheads independent of network hardware.  Using point-to-point 
 protocol details, we explore the decomposition of MPI time into communicat
 ion, synchronization and software stack components using architecture simu
 lation.  Detailed validation using Bayesian inference is used to identify 
 the sensitivity of performance to specific latency/bandwidth parameters fo
 r different network protocols and to quantify associated uncertainties. Th
 e inference combined with trace replay shows that synchronization and MPI 
 software stack overhead are at least as important as the network itself in
  determining time spent in communication routines.\n\nPasses: Conference P
 ass, MPI, Networks, Performance Analysis and Optimization, Performance Too
 ls\n\nTag: Conference Pass, MPI, Networks, Performance Analysis and Optimi
 zation, Performance Tools
URL:https://2018.isc-program.com/presentation/?id=pap115&sess=sess161
END:VEVENT
END:VCALENDAR