Course Content:
In this course the participants will learn about modern and advanced concepts of MPI and will be able to use those features in their application, with the goal to increase robustness and performance.
The course will discuss (all subject to change):
- Persistent and partitioned communication
- Virtual topologies and neighborhood collectives
- MPI RMA and shared memory
- Intercommunicators
- MPI Sessions
- The MPI Tools Interface
Prerequisites:
Participants should have experience with the basic concepts of MPI, in particular point-to-point and collective communication
Target audience:
Scientists, PhD students, postdocs, supercomputer users, programmers who want to use advanced features of MPI
Language:
This course is given in English.
Duration:
3,5 days
Dates:
30.11.-03.12.2026, 09:00-17:00h (12:00 on Thursday)
Venue:
Jülich Supercomputing Centre, building 16.3, Training Room 1 (room 213a)
Number of Participants:
Minimum: 5; Maximum: 30
Instructor:
Michael Knobloch (JSC)
Fees
Free of charge
Pre-required logistics
Participants should have experience with the basic concepts of MPI, in particular point-to-point and collective communication