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:20230124T171523Z
LOCATION:C1-2-3
DTSTART;TZID=America/Chicago:20221117T083000
DTEND;TZID=America/Chicago:20221117T170000
UID:submissions.supercomputing.org_SC22_sess275_rpost104@linklings.com
SUMMARY:Learning to Parallelize Source Code via OpenMP with Transformers
DESCRIPTION:Posters, Research Posters\n\nLearning to Parallelize Source Co
 de via OpenMP with Transformers\n\nHarel, Pinter, Oren\n\nIn past years, t
 he world has switched to many-core and multi-core shared memory architectu
 res. As a result, there is a growing need to utilize these architectures b
 y introducing shared memory parallelization schemes, such as OpenMP, to so
 ftware applications. Nevertheless, introducing OpenMP into code, especiall
 y legacy code, is challenging due to pervasive pitfalls in management of p
 arallel shared memory.  To facilitate the performance of this task, many s
 ource-to-source (S2S) compilers have been created over the years, tasked w
 ith inserting OpenMP directives into code automatically.  In addition to h
 aving limited robustness to their input format, these compilers still do n
 ot achieve satisfactory coverage and precision in locating parallelizable 
 code and generating appropriate directives.  In this work, we propose leve
 raging recent advances in machine learning techniques, specifically in nat
 ural language processing (NLP) - the transformers model, to suggest the ne
 ed for an OpenMP directive or specific clauses (reduction and private).\n\
 nRegistration Category: Tech Program Reg Pass, Exhibits Reg Pass
END:VEVENT
END:VCALENDAR