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 Advanced Training Center event. It is organized by LRZ in cooperation with HLRS, RRZE, and VSC (Vienna Scientific Cluster).
10:45 Programming Models
- Pure MPI
11:05 Coffee Break
11:25 - Topology Optimization
12:05 Practical (application aware Cartesian topology)
12:45 - Topology Optimization (Wrap up)
14:00 - MPI + MPI-3.0 Shared Memory
14:30 Practical (replicated data)
15:00 Coffee Break
15:20 - MPI Memory Models and Synchronization
16:00 Practical (substituting pt-to-pt by shared memory)
16:45 Coffee Break
17:00 Practical (substituting barrier synchronization by pt-to-pt)
Social Event (self paying)
09:00 Programming Models (continued)
- MPI + OpenMP
10:30 Coffee Break
10:50 Practical (how to compile and start)
11:30 Practical (hybrid through OpenMP parallelization)
14:00 - Overlapping Communication and Computation
14:20 Practical (taskloops)
15:00 Coffee Break
15:20 - MPI + OpenMP Conclusions
15:30 - MPI + Accelerators
Basic MPI and OpenMP knowledge as presented in the LRZ course "Parallel programming of High Performance Systems". For the hands-on sessions you should know Unix/Linux and either C/C++ or Fortran in particular.
The course language is English.
Registration via https://events.prace-ri.eu/event/807/registration/register
For further information, see also course pages
for registration is Jan. 14, 2019.
This course is a PRACE Advanced Training Center event. Therefore, the course is free of charge for all participants from the EU or from PRACE-member-countries.
Volker Weinberg phone 089 35831 8863, weinberglrz.de
http://vsc.ac.at/training (at VSC, Vienna Scientific Cluster)