BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
BEGIN:VEVENT
DTSTAMP:20260704T160721Z
UID:4cc2bccb-f8b6-48ba-ace5-6bc5da82d9ad
DTSTART:20161213T080000Z
DTEND:20161214T160000Z
DESCRIPTION:The purpose of this training course is to present hybrid progra
 mming\, MPI+OpenMP\, as well as to give a feedback of effective implementa
 tions of such a parallel programming model on several application codes.We
  will explain why the hybrid programming approach seems to be the best sui
 ted model for the next massively parallel architectures. Indeed\, the bene
 fits of hybrid programming are numerous: memory savings\, better performan
 ces\, better load balancing\, greater granularity\, improved scalability\,
  better adequacy of the code to the hardware specificities of the target a
 rchitecture.Content:\n\nIntroduction to hybrid MPI-OpenMP parallelization\
 ,\n	Hybrid MPI-OpenMP parallelization concepts\,\n	Feedback on several rea
 l applications\,\n	Hands-on sessions will use IDRIS supercomputers\nLearni
 ng outcome:After this course\, participants will be able to:\n\nOptimize t
 he original parallel OpenMP and MPI versions of their code\n	Choose the be
 st strategy\, especially the level of thread support\, depending on their 
 code and on the target architecture characteristics\n	Implement the strate
 gy to obtain an efficient parallel hybrid MPI+OpenMP version.\nThey will a
 lso know how to use performance/debugging tools to find bugs\, deadlocks o
 r determine the part of their code that leads to poor performance.Pre-requ
 isites:Advanced users who have a good knowledge of: C or Fortran programmi
 ng language\, MPI and OpenMP~\n\nhttps://events.prace-ri.eu/event/555/
SUMMARY:Hybrid MPI/OpenMP programming @ MdlS/IDRIS
URL;VALUE=URI:https://events.prace-ri.eu/event/555/
END:VEVENT
END:VCALENDAR