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:20210402T160552Z
LOCATION:Track 1
DTSTART;TZID=America/New_York:20201113T120000
DTEND;TZID=America/New_York:20201113T123000
UID:submissions.supercomputing.org_SC20_sess220_ws_exampi108@linklings.com
SUMMARY:Scalable MPI Collectives Using SHARP: Large Scale Performance Eval
 uation on the TACC Frontera System
DESCRIPTION:Workshop\n\nScalable MPI Collectives Using SHARP: Large Scale 
 Performance Evaluation on the TACC Frontera System\n\nRamesh, Kandadi Sure
 sh, Sarkauskas, Bayatpour, Maqbool Hashmi...\n\nMessage-Passing Interface 
 (MPI) is the de-facto standard for designing and executing applications on
  massively parallel hardware. MPI collectives provide a convenient abstrac
 tion for multiple processes/threads to communicate with one another. Mella
 nox’s HDR InfiniBand switches provide Scalable Hierarchical Aggregation an
 d Reduction Protocol (SHARP) capabilities to offload collective communicat
 ion to the network and reduce CPU involvement in the process. In this pape
 r, we propose, design and implement SHARP-based solutions for MPI Reduce a
 nd MPI Barrier in MVAPICH2-X. We evaluate the impact of proposed and exist
 ing SHARP-based solutions for MPI Allreduce, MPI Reduce and MPI Barrier op
 erations on the performance of the collective operation on the 8th ranked 
 TACC Frontera HPC system. Our experimental evaluation of the SHARP-based d
 esigns shows up to 5.4x reduction in latency for Reduce, 5.1x for Allreduc
 e and 7.1x for Barrier at full system scale of 7,861 nodes over a host-bas
 ed solution.\n\nRegistration Category: Workshop Reg Pass
END:VEVENT
END:VCALENDAR