BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240116T191700Z
LOCATION:601
DTSTART;TZID=America/Denver:20231113T103000
DTEND;TZID=America/Denver:20231113T105000
UID:submissions.supercomputing.org_SC23_sess443_ws_ai4dev102@linklings.com
SUMMARY:MPI-RICAL: Data-Driven MPI Distributed Parallelism Assistance with
  Transformers
DESCRIPTION:Workshop\n\nNadav Schneider and Tal Kadosh (Ben-Gurion Univers
 ity of the Negev, Israel; IAEC); Niranjan Hasabnis and Timothy Mattson (In
 tel Labs); Yuval Pinter (Ben-Gurion University of the Negev, Israel); and 
 Gal Oren (Technion - Israel Institute of Technology)\n\nMessage Passing In
 terface (MPI) plays a crucial role in distributed memory parallelization a
 cross multiple nodes. However, parallelizing MPI code manually, and specif
 ically, performing domain decomposition, is a challenging, error-prone tas
 k. In this paper, we address this problem by developing MPI-RICAL, a novel
  data-driven, programming-assistance tool that assists programmers in writ
 ing domain decomposition based distributed memory parallelization code. Sp
 ecifically, we train a supervised language model to suggest MPI functions 
 and their proper locations in the code on the fly. We also introduce MPICo
 deCorpus, the first publicly available corpus of MPI-based parallel progra
 ms that is created by mining more than 15,000 open-source repositories on 
 GitHub. Experimental results have been done on MPICodeCorpus and more impo
 rtantly, on a compiled benchmark of MPI-based parallel programs for numeri
 cal computations that represent real-world scientific applications. MPI-RI
 CAL achieves F1 scores between 0.87-0.91 on these programs, demonstrating 
 its accuracy in suggesting correct MPI functions at appropriate code locat
 ions.\n\nTag: Artificial Intelligence/Machine Learning, Software Engineeri
 ng\n\nRegistration Category: Workshop Reg Pass\n\nSession Chairs: Giorgis 
 Georgakoudis (Lawrence Livermore National Laboratory (LLNL)), Ignacio Lagu
 na (Lawrence Livermore National Laboratory (LLNL)), and Konstantinos Paras
 yris (Lawrence Livermore National Laboratory (LLNL))
END:VEVENT
END:VCALENDAR