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:20260522T150111Z
LOCATION:C2/3/4 Ballroom
DTSTART;TZID=America/Chicago:20181115T083000
DTEND;TZID=America/Chicago:20181115T170000
UID:submissions.supercomputing.org_SC18_sess324_post250@linklings.com
SUMMARY:Exploring Application Performance on Fat-Tree Networks in the Pres
 ence of Congestion
DESCRIPTION:Philip A. Taffet (Rice University, Lawrence Livermore National
  Laboratory); Sanil Rao (University of Virginia, Lawrence Livermore Nation
 al Laboratory); and Ian Karlin (Lawrence Livermore National Laboratory)\n\
 nNetwork congestion, which occurs when multiple applications simultaneousl
 y use shared links in cluster network, can cause poor communication perfor
 mance, decreasing the performance and scalability of parallel applications
 . Many studies are performed while clusters also run other production work
 loads, which makes it harder for them to isolate causes and their effects.
  To look at congestion in a more controlled setting we used dedicated acce
 ss time on an HPC cluster and measured the performance of three HPC applic
 ations with different communication patterns run with varying amounts and 
 types of background traffic. This enables us to assess the relative sensit
 ivity of the applications to congestion caused by different traffic patter
 ns. Our tests show that the applications were not significantly impacted b
 y even the most aggressive neighboring patterns, with all the performance 
 degradation being 7% or less, pointing to the resiliency of the fat-tree t
 opology.\n\nRegistration Category: Tech Program Reg Pass, Exhibits Reg Pas
 s\n\n
END:VEVENT
END:VCALENDAR