Parallel processing for co-simulation tends to be more involved than it is for single-solver simulations. A co-simulation involves, at minimum, three participants (that is, System Coupling and at least two coupling participants). Figure 28: Co-simulation with four coupling participants shows a co-simulation with four participants, each with different number of compute processes per participant. System Coupling provides participant partitioning capabilities, partitioning the simulation work across participants then allocating that work across available computing resources.
When setting up participants for parallel execution, you can set the following parameters to control the behavior of the run:
Machine List: The list of machines to be used and, where applicable, the number of cores each has available for the parallel run. For more information, see Machine List and Core Counts.
Participant Fractions: The fractions specifying the core count or fraction of compute resources to be allocated to each participant. For more information, see Resource Allocation Fractions.
Partitioning Algorithm: The algorithm used to specify the resource allocation method (machines or cores) and parallel execution mode (local or distributed) to be used. For more information, see Partitioning Algorithms.
The parallel execution details that were specified most recently — regardless of by what method — replace existing parallel values.
For more information, see: