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:20260522T150119Z
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:Hadia Ahmed and Scott B. Baden (Lawrence Berkeley National Lab
 oratory)\n\nThe continuous increase in the number of nodes across the whol
 e system in supercomputers makes Remote-memory-access (RMA) a good alterna
 tive for improving applications' performance on these machines. However, i
 n order for existing applications to benefit from it, they have to change 
 their code to introduce the RMA functionalities. Doing so manually is erro
 r-prone.\n\nIn our research, we are working to develop a translator tool t
 o generate UPC++ code that has RMA functionality. UPC++ is a C++11 library
  providing classes and functions that includes RMA communication operation
 s. \n\nIn our first prototype, the translator uses annotations to guide tr
 ansformation. First, it abstracts the message as a global pointer and enab
 les ranks to distribute their local copies of their messages to other rank
 s. Then, it translates the MPI two-sided communication to corresponding UP
 C++ (RMA-rput) and a local synchronization. As a proof of concept, we appl
 ied this translator to stencil-method solver applications. The translator 
 successfully generated a correct UPC++ code that achieves similar performa
 nce to the MPI version. \n\nWith stencil applications we do not expect bet
 ter performance from corresponding MPI version. However, we were able to d
 emonstrate that we can produce correct UPC++ implementation. Our next step
  is to apply the translator on applications that demonstrates irregular co
 mmunication and/or fine-grained messaged. These are expected to show bette
 r performance with UPC++ under distributed memory architectures. The overa
 ll goal of this work is to help scientist in adapting their code to use PG
 AS effortlessly. Hence, they benefit from HPC resources and achieve better
  performance.\n\nTag: Diversity, Education, Hot Topics\n\nRegistration Cat
 egory: Workshop Reg Pass\n\nSession Chairs: Toni Collis (Women in High Per
 formance Computing); Weronika Filinger (Edinburgh Parallel Computing Centr
 e (EPCC); University of Edinburgh, Scotland); and Misbah Mubarak (Amazon W
 eb Services)\n\n
END:VEVENT
END:VCALENDAR