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:20181221T160726Z
LOCATION:D220
DTSTART;TZID=America/Chicago:20181111T143900
DTEND;TZID=America/Chicago:20181111T144200
UID:submissions.supercomputing.org_SC18_sess160_ws_whpc119@linklings.com
SUMMARY:From Message Passing to PGAS
DESCRIPTION:Workshop\nDiversity, Education, Hot Topics, Workshop Reg Pass\
 n\nFrom Message Passing to PGAS\n\nAhmed, Baden\n\nThe continuous increase
  in the number of nodes across the whole system in supercomputers makes Re
 mote-memory-access (RMA) a good alternative for improving applications' pe
 rformance on these machines. However, in order for existing applications t
 o benefit from it, they have to change their code to introduce the RMA fun
 ctionalities. Doing so manually is error-prone.\n\nIn our research, we are
  working to develop a translator tool to generate UPC++ code that has RMA 
 functionality. UPC++ is a C++11 library providing classes and functions th
 at includes RMA communication operations. \n\nIn our first prototype, the 
 translator uses annotations to guide transformation. First, it abstracts t
 he message as a global pointer and enables ranks to distribute their local
  copies of their messages to other ranks. Then, it translates the MPI two-
 sided communication to corresponding UPC++ (RMA-rput) and a local synchron
 ization. As a proof of concept, we applied this translator to stencil-meth
 od solver applications. The translator successfully generated a correct UP
 C++ code that achieves similar performance to the MPI version. \n\nWith st
 encil applications we do not expect better performance from corresponding 
 MPI version. However, we were able to demonstrate that we can produce corr
 ect UPC++ implementation. Our next step is to apply the translator on appl
 ications that demonstrates irregular communication and/or fine-grained mes
 saged. These are expected to show better performance with UPC++ under dist
 ributed memory architectures. The overall goal of this work is to help sci
 entist in adapting their code to use PGAS effortlessly. Hence, they benefi
 t from HPC resources and achieve better performance.
URL:https://sc18.supercomputing.org/presentation/?id=ws_whpc119&sess=sess1
 60
END:VEVENT
END:VCALENDAR