BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
BEGIN:VEVENT
DTSTAMP:20260704T012514Z
UID:979840cb-e788-4e09-85f4-5a709f9debd9
DTSTART:20160606T073000Z
DTEND:20160607T153000Z
DESCRIPTION:Keywords\nRuntime systems\, thread programming\, accelerators\,
  hardware abstraction\, performance portability.\n \n\nCourse description
 :\nThis course will present the state of the art of runtime system support
  for programming heterogeneous platforms. Heterogeneous computing platform
 s—such as multicores equipped with accelerators—are notoriously diffic
 ult to program due to the strong differences in performance characteristic
 s among the various available computing units and also to the discrete mem
 ory spaces of accelerating boards.\nThe course will present the StarPU run
 time system developed at Inria by the STORM Team in Bordeaux. It will also
  present the hardware locality library hwloc for discovering hardware reso
 urces and the TreeMatch framework (Bordeaux / Tadaam) for distributed proc
 esses placement\, and the Ezperf framework for solving performance issues 
 (SED Bordeaux). \n\nLearning outcomes\nParticipants will understand the be
 nefit of the task-based programming model together with performance modeli
 ng\, automatic data management and dynamic scheduling in speeding up appli
 cation development on heterogeneous computing platforms and providing long
  term performance portability.Prerequisites\nBasic knowledge of C programm
 ing language and accelerator programming languages (nVidia Cuda\, OpenCL).
  \n\nhttps://events.prace-ri.eu/event/453/
SUMMARY:Runtime systems for heterogeneous platform programming@MDS
URL;VALUE=URI:https://events.prace-ri.eu/event/453/
END:VEVENT
END:VCALENDAR