BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
BEGIN:VEVENT
DTSTAMP:20260703T152447Z
UID:cef5ef48-6403-44d9-944c-f437b17658cd
DTSTART:20220111T093000Z
DTEND:20220111T170000Z
DESCRIPTION:One of the greatest challenges to running parallel applications
  on\nlarge numbers of processors is how to handle file IO. Standard Unix I
 O\nroutines are not designed with parallelism in mind\, and IO overheads\n
 can grow to dominate the overall runtime. Parallel file systems are\noptim
 ised for large volumes of data\, but performance can be far from\noptimal 
 if every process opens its own file or if all IO is funnelled\nthrough a s
 ingle controller process.\n\nThis hands-on course explores a range of issu
 es related to parallel\nIO. It uses ARCHER2 and its parallel Lustre file s
 ystem as a platform\nfor the exercises\; however\, almost all the IO conce
 pts and performance\nconsiderations are applicable to any parallel system.
 \n\nWe will give a general overview of how parallel IO is implemented in\n
 MPI-IO as these are the routines ultimately used by higher-level\nlibrarie
 s such as HDF5 and NetCDF. A good understanding of the\nperformance charac
 teristics of MPI-IO is therefore very useful in\noptimising the IO perform
 ance of most parallel applications.\n\nPrerequisites: The course assumes a
  good understanding of basic MPI\nprogramming in C\, C++ or Fortran. Knowl
 edge of MPI derived datatypes\nwould be useful but not essential.\nhttps:/
 /events.prace-ri.eu/event/1299/
SUMMARY:[ONLINE] Efficient Parallel IO on ARCHER2 @ EPCC  11 Jan 2022
URL;VALUE=URI:https://events.prace-ri.eu/event/1299/
END:VEVENT
END:VCALENDAR