Programming Parallel Computers
- Marc-André Hermanns (Research Centre Jülich, Germany)
- Bernd Mohr (Research Centre Juelich, Germany)
- Wolfgang Frings (Research Centre Jülich, Germany)
This tutorial gives you a thorough introduction to scientific parallel programming. After an overview of parallel computing, algorithms, and programming, in the main part of the course you learn how to write parallel programs using the standard Message Passing Interface (MPI). The third part of the tutorial introduces you to shared memory programming with OpenMP.
Lectures alternate with hands-on exercises. The exercises provide an excellent opportunity to apply the techniques just presented. Attendees will receive copies of the lectures and additional reference material on MPI and OpenMP. There will also be time for discussions with scientists from various fields of computational science, including parallel numerical computing, quantum chemistry, and molecular dynamics.
* Good knowledge of a sequential programming language (e.g., C, Fortran, C++)
* Experience in scientific and numerical programming
[Day 1]Introduction to Parallel Programming
* Parallel computer architectures and computer systems
* Parallel algorithms and software
* Parallel programming
[Day 2-4] Introduction to MPI Programming
* Basic concepts
* Point-to-point message passing
* Collective communication
* Advanced topics
[Day 5] Introduction to OpenMP Programming
Free lodging is provided to all tutorial participants sharing 2 person rooms in a Jülich downtown hotel.
Free breakfast will be provided in the hotel and free lunches will be provided at the tutorial site. The participants are responsible for dinners themselves.
There will be a free shuttle bus service between the hotels and the tutorial site.
Participants are responsible for the travel to and from Jülich. There will be partial travel support through CECAM. Accepted participants are requested to fill in the forms with their personal data.
Participants are invited to bring a poster describing their research work.