Blockkurs zu Parallel Programming with C and MPI

• Termin: 19.05.2008 - 12.06.2008, 15 Doppelstunden
• Dozent: Prof. Dr. G. S. Adhar
• Abstract:
  In this course students will learn to write parallel programs using the MPI (Message Passing Interface) on a cluster of computers. The goal of this course is to write high performance code using the MPI application programming interface which has become de-facto standard today in the high performance world.
  
  Starting with the MPI support for the point-to-point communication followed by discussion of the group communication functions the course will examine MPI support for user defined communicators for grids and the user defined data types.
  
  MPI extensions for logging events will be presented to help with probing of parallel programs and profiling tools such as jumpshot will be used to visualize the event logs to fine tune the parallel programs.
  
  Topics will be presented within the context of specific applications drawn from many areas in computer science such as: Simulated Annealing, Ant-Colony Optimization, Matrix Computations, and Combinatorial Optimization.
  
  Prerequisite: C programming
  
  
• Course, room and times:
  
  Mo - Fr, 10.00 - 12.00
  CAP4 - R.709
  
  
• First meeting, room and time:
  
  Mo, 14.00
  CAP4 - R.910 [SR910]
  
  
• Ansprechpartner: Jaroslaw Piwonski, Raum 917.