Parallel programming with MPI (Hybrid Course)

Topics covered

The Message Passing Interface is the de facto standard for distributed programming in a scientific context. It is implemented in libraries that can be  used from C/C++/Fortran, although wrappers for other languages exist as well. To write efficient code, a thorough understanding of the semantics of the API is required, this course will emphasize potential performance hazards and pitfalls.  The strengths and weaknesses of various MPI features are discussed. In several hands-on sessions, the participant has the opportunity to experiment.

• Parallel architectures and programming models
• MPI overview
• MPI programming model
• Messages and point-to-point communication
• Non-blocking communication
• Derived data types
• Virtual topologies
• Collective communication
• MPI-2 overview
• MPI-2 one-sided communication
• Parallel debugging
• MPI-2 parallel file
• MPI-3 overview

This MPI-course covers the full MPI-3 standard.

Target audience

This info session is primarily targeted at VSC-users, although other interested parties are welcome as well.

Previous knowledge

A good working knowledge of C/C++ or Fortran programming is expected. Basic knowledge of Linux usage and HPC systems helps. Level: Intermediate

Result/Objectives

• Participants are able to start developing distributed applications using MPI
• Good understanding of the semantics of MPI features
• Awareness of potential pitfalls
• Familiarity with the new features in the MPI-3 standard

Remark The course material was developed by dr. Rolf Rabenseifner (HRLS, Stuttgart, Germany).