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:20210402T160053Z
LOCATION:Track 5
DTSTART;TZID=America/New_York:20201117T153000
DTEND;TZID=America/New_York:20201117T160000
UID:submissions.supercomputing.org_SC20_sess161_pap447@linklings.com
SUMMARY:An In-Depth Analysis of the Slingshot Interconnect
DESCRIPTION:Paper\n\nAn In-Depth Analysis of the Slingshot Interconnect\n\
 nDe Sensi, Di Girolamo, McMahon, Roweth, Hoefler\n\nThe interconnection ne
 twork is one of the most critical components in large scale computing syst
 ems, and its impact on the performance of applications is going to increas
 e with the system size. In this paper, we describe Slingshot, an interconn
 ection network for large scale computing systems. Slingshot is based on hi
 gh-radix switches, which allow building exascale and hyperscale datacenter
  networks with at most three switch-to-switch hops. Moreover, Slingshot pr
 ovides efficient adaptive routing and congestion control mechanisms, and h
 ighly tunable traffic classes. Slingshot uses an optimized Ethernet protoc
 ol, which allows it to be interoperable with standard Ethernet devices whi
 le providing high performance to HPC applications. We analyze the extent t
 o which Slingshot provides these features, evaluating it with microbenchma
 rks and with several applications from the datacenter and AI worlds, as we
 ll as with HPC applications. We find that applications running on Slingsho
 t are less affected by congestion compared to previous generation networks
 .\n\nTag: Algorithms, MPI, Networks\n\nRegistration Category: Tech Program
  Reg Pass
END:VEVENT
END:VCALENDAR