How can I use the Parallel Computing Toolbox to run multiple Simulink simulations in a parallel or distributed fashion?

I would like to know whether it is possible to use the Parallel Computing Toolbox to do either of the following:

1) Run multiple instances of one Simulink model on different cores, processors, or cluster machines. The instances would be running according to different, independent sets of parameters, or operating on different, independent data sets.

2) Run different Simulink models on different cores, processors, or cluster machines.

Answer 1

[MathWorks Support Team]

1. Running a single simulation on multiple cores

A single Simulink simulation is primarily single-threaded, but specific parts of the simulation can execute in parallel when the model and configuration allow. Key mechanisms include:

Many built-in Simulink blocks (e.g., matrix operations, linear algebra) rely on multithreaded BLAS/LAPACK libraries. These automatically use multiple CPU cores when performing large computations.

LAPACK in MATLAB - MATLAB & Simulink

The Dataflow domain enables concurrent execution of independent parts of a model using a data-driven execution model.

Dataflow Domain - MATLAB & Simulink

For Each Subsystems can be configured to process elements of an array or signal in parallel.

For Each - Control block for for-each subsystem - Simulink

Certain co-simulation workflows (e.g., FMU import, external solvers, or model references in accelerator modes) can execute concurrently depending on solver and partitioning setup.

Multithread Co-Simulation - MATLAB & Simulink 

Summary:

You cannot generally parallelize an entire single simulation automatically, but there are limited features for parallel execution in specific parts of the model using Dataflow, For Each subsystems, and multithreaded libraries.

2. Running multiple simulations in parallel (recommended approach)

If your goal is to sweep parameters, run test cases, or evaluate scenarios, the recommended modern approach is to run independent simulations in parallel.

Use parsim to run multiple simulations concurrently in a parallel pool.

in(1:10) = Simulink.SimulationInput("myModel");

for i = 1:10

    in(i) = in(i).setVariable("K", i);

end

out = parsim(in);

Key benefits:

• Easy parameter sweeps
• Works with Parallel Computing Toolbox
• Supports fast restart and model reuse

Use batchsim when you want to run simulations:

• On a cluster
• In the background

Summary:

For most workflows (parameter sweeps, Monte Carlo, regression testing), parallel simulations using parsim or batchsim are the preferred and most scalable solution.

[Walter Roberson]

From R2017a onward, parsim was introduced. See https://blogs.mathworks.com/simulink/2016/05/05/tips-for-simulating-models-in-parallel/