Most HPC systems are clusters of shared memory nodes. Such SMP nodes can be small multi-core CPUs up to large many-core CPUs. Parallel programming may combine the distributed memory parallelization on the node interconnect (e.g., with MPI) with the shared memory parallelization inside of each node (e.g., with OpenMP or MPI-3.0 shared memory). This course analyzes the strengths and weaknesses of several parallel programming models on clusters of SMP nodes. Multi-socket-multi-core systems in highly parallel environments are given special consideration. MPI-3.0 has introduced a new shared memory programming interface, which can be combined with inter-node MPI communication. It can be used for direct neighbor accesses similar to OpenMP or for direct halo copies, and enables new hybrid programming models. These models are compared with various hybrid MPI+OpenMP approaches and pure MPI. Numerous case studies and micro-benchmarks demonstrate the performance-related aspects of hybrid programming.

Hands-on sessions are included on both days. Tools for hybrid programming such as thread/process placement support and performance analysis are presented in a "how-to" section. This course provides scientific training in Computational Science, and in addition, the scientific exchange of the participants among themselves. The course is a PRACE training event. It is organized by LRZ in cooperation with HLRS, RRZE, and VSC (Vienna Scientific Cluster).

The materials of this annual PRACE course can be found here: https://moodle.rrze.uni-erlangen.de/course/view.php?id=388 (onsite version 2019) and https://moodle.rrze.uni-erlangen.de/course/view.php?id=399 (latest online version 2020).


Dr. habil. Georg Hager (RRZE, Uni. Erlangen), Dr. Rolf Rabenseifner (Stuttgart), Dr. Claudia Blaas-Schenner and Dr. Irene Reichl (VSC Team, TU Wien)


The online version of the course has been extended to three days (as compared to the originally planned two days face-to-face course) to allow for more time and flexibility with doing the hands-on labs remotely.

1st day – 17 June 2020

08:45   Join online
09:00    Welcome
09:05    Motivation
09:15    Introduction
09:45    Programming Models
09:50       - MPI + OpenMP
10:30   Coffee
11:00       - continue: MPI + OpenMP
12:00          Practical (how to compile and start)
13:00   Lunch
14:30          Practical (hybrid through OpenMP parallelization)
16:00   End of first day

2nd day – 18 June 2020

08:45   Join online
09:00       - Overlapping Communication and Computation
09:30          Practical (taskloops)
10:30   Coffee
11:00       - MPI + OpenMP Conclusions
11:15       - MPI + Accelerators
11:45    Tools
12:00    Programming Models (continued)
12:05       - MPI + MPI-3.0 Shared Memory
13:00   Lunch
14:30          Practical (replicated data)
16:00   End of second day

3rd day – 19 June 2020

08:45   Join online
09:00       - MPI Memory Models and Synchronization
09:40       - Pure MPI
10:00       - Recap - MPI Virtual Topologies
10:30   Coffee
11:00       - Topology Optimization
11:30          Practical/Demo (application aware Cartesian topology)
12:30       - Topology Optimization (Wrap up)
12:45     Conclusions
13:00   Lunch
14:30     Finish the hands-on labs, Discussion, Q & A, Feedback
16:00   End of third day (course)

  • No labels