BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
BEGIN:VEVENT
DTSTAMP:20260704T184147Z
UID:38612614-88d6-45f3-8ea9-c66e15ba1aa4
DTSTART:20170329T080000Z
DTEND:20170330T163000Z
DESCRIPTION:Please note:  This course takes place in Durham.\n\nOne of the
  greatest challenges to running parallel applications on\nlarge numbers of
  processors is how to handle file IO. Standard IO\nroutines are not design
 ed with parallelism in mind\, and IO overheads\ncan grow to dominate the o
 verall runtime. Parallel file systems are\noptimised for large data transf
 ers\, but performance can be far from\noptimal if every process opens its 
 own file or if all IO is funneled\nthrough a single master process.\n\nThi
 s hands-on course explores a range of issues related to parallel\nIO. It u
 ses ARCHER and its parallel Lustre file system as a platform\nfor the exer
 cises\; however\, almost all the IO concepts and performance\nconsideratio
 ns are applicable to any parallel system.\n\nThe IO part of the MPI standa
 rd gives programmers access to efficient\nparallel IO in a portable fashio
 n. However\, there are a large number\nof different routines available and
  some can be difficult to use in\npractice. Despite its apparent complexit
 y\, MPI-IO adopts a very\nstraightforward high-level model. If used correc
 tly\, almost all the\ncomplexities of aggregating data from multiple proce
 sses can be dealt\nwith automatically by the library.\n\nThe first day of 
 the course will cover the MPI-IO standard\, developing\nIO routines for a 
 regular domain decomposition example. It will also\nbriefly cover higher-l
 evel standards such as HDF5 and NetCDF which are\nbuilt on top of MPI-IO.\
 n\nThe second day will concentrate on performance\, covering how to\nconfi
 gure the parallel file system and tune the MPI-IO library for\nbest perfor
 mance. Case studies from real codes will be presented.\n\nPrerequisites: T
 he course assumes a good understanding of basic MPI\nprogramming in Fortra
 n\, C or C++. Knowledge of MPI derived datatypes\nwould be useful but not 
 essential.\n\nTimetable\n\nWednesday 29 March\n\n09:30 - 10:15 : Parallel 
 IO\n10:15 - 11:00 : Practical : Basic IO\n11:00 - 11:30 : Break\n11:30 - 1
 2:15 : Derived Datatypes for MPI-IO\n12:15 - 13:00 : Practical: Derived Da
 tatypes\n13:00 - 14:00 : Lunch\n14:00 - 14:45 : Basic MPI-IO Routines\n14:
 45 - 15:30 : Practical: Basic MPI-IO\n15:30 - 16:00 : Break\n16:00 - 16:45
  : MPI-IO Features and alternative libraries\n16:45 - 17:30 : Practical : 
 Alternative Libraries\n\nThursday 30th March\n\n09:30 - 10:15 : Lustre fil
 e 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 parallel IO\n13:00 - 14:00 : Lunch\n14:00 - 14:
 45 : Case studies\n14:45 - 15:30 : Individual consultancy session\n\nhttps
 ://events.prace-ri.eu/event/591/
SUMMARY:Efficient Parallel IO on ARCHER 29-30 Mar 2017 @ EPCC at Durham
URL;VALUE=URI:https://events.prace-ri.eu/event/591/
END:VEVENT
END:VCALENDAR