BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
BEGIN:VEVENT
DTSTAMP:20260708T120916Z
UID:92278ea8-f688-4ab7-82be-ec2a7d160f71
DTSTART:20190328T090000Z
DTEND:20190329T173000Z
DESCRIPTION:Please note:  This course will take place at Imperial College 
 London\n\nSALC 10\,  Sherfield building\, level 5\, South Kensington Camp
 us\n\nOne of the greatest challenges to running parallel applications on l
 arge numbers of processors is how to handle file IO. Standard IO routines 
 are not designed with parallelism in mind\, and IO overheads can grow to d
 ominate the overall runtime. Parallel file systems are optimised for large
  data transfers\, but performance can be far from optimal if every process
  opens its own file or if all IO is funneled through a single master proce
 ss.\n\nThis hands-on course explores a range of issues related to parallel
  IO. It uses ARCHER and its parallel Lustre file system as a platform for 
 the exercises\; however\, almost all the IO concepts and performance consi
 derations are applicable to any parallel system.\n\nThe IO part of the MPI
  standard gives programmers access to efficient parallel IO in a portable 
 fashion. However\, there are a large number of different routines availabl
 e and some can be difficult to use in practice. Despite its apparent compl
 exity\, MPI-IO adopts a very straightforward high-level model. If used cor
 rectly\, almost all the complexities of aggregating data from multiple pro
 cesses can be dealt with automatically by the library.\n\nThe first day of
  the course will cover the MPI-IO standard\, developing IO routines for a 
 regular domain decomposition example. It will also briefly cover higher-le
 vel standards such as HDF5 and NetCDF which are built on top of MPI-IO.\n\
 nThe second day will concentrate on performance\, covering how to configur
 e the parallel file system and tune the MPI-IO library for best performanc
 e. Case studies from real codes will be presented.\n\nPrerequisites: The c
 ourse assumes a good understanding of basic MPI programming in Fortran\, C
  or C++. Knowledge of MPI derived datatypes would be useful but not essent
 ial.\n\nTimetable\n\nDay 1\n\n09:30 - 10:15 : Parallel IO\n10:15 - 11:00 :
  Practical : Basic IO\n11:00 - 11:30 : Break\n11:30 - 12:15 : Derived Data
 types for MPI-IO\n12:15 - 13:00 : Practical: Derived Datatypes\n13:00 - 14
 :00 : Lunch\n14:00 - 14:45 : Basic MPI-IO Routines\n14:45 - 15:30 : Practi
 cal: Basic MPI-IO\n15:30 - 16:00 : Break\n16:00 - 16:45 : MPI-IO Features 
 and alternative libraries\n16:45 - 17:30 : Practical : Alternative Librari
 es\n\nDay 2\n\n09:30 - 10:15 : Lustre file system on ARCHER\n10:15 - 11:00
  : Practical: Lustre configuration\n11:00 - 11:30 : Break\n11:30 - 12:15 :
  Parallel IO libraries on ARCHER\n12:15 - 13:00 : Practical: tuning parall
 el IO\n13:00 - 14:00 : Lunch\n14:00 - 14:45 : Case studies\n14:45 - 15:30 
 : Individual consultancy session\n\n \n\nTrainer\n\nDavid Henty\n\nDavid 
 teaches on a wide range of EPCC's technical training courses\, including M
 PI and OpenMP\, and is overall course organiser for EPCC's MSc in High Per
 formance Computing.\n\n \n\n \n\nCourse Material\n\nhttp://www.archer.ac
 .uk/training/course-material/2019/03/io-imp/index.php \nhttps://events.pra
 ce-ri.eu/event/846/
SUMMARY:Efficient Parallel I/O @ EPCC at Imperial College London
URL;VALUE=URI:https://events.prace-ri.eu/event/846/
END:VEVENT
END:VCALENDAR