Using parallel computing WITHIN a Simulink simulation

I have a for loop within a my simulink model (using the For Iterator block). In my case, I have around 2000-3000 iterations each time step. From my searching, I have only found that parallel computing can only be used to run simulations in parallel, rather than within a single simulation. Is it possible to use parallel computing for the for loop within a single simulink simulation?

Answer 1

[Ryan G]

You cannot use parallel computing for this operation. Can I ask why you are doing this? Although there may be a perfectly valid reason, perhaps reconsidering your approach may help speed things up.

[michaelborg]

I have a number of input components (around 1000), and for each component I have to use a lookup table to find a corresponding value. I wish to use parallel computing because initially I initially had 50 input components, but when I increased this to 1000 to get more accurate results, the simulation significantly slowed down. Is it possible to use the parfor statement in a Matlab Function block?

[Ryan G]

I don't believe you can do it in a MATLAB function either. Maybe, but I wouldn't try it.

Have you looked into turning the subsystem into a model reference and running in accelerator mode? That would definitely speed things up and is supported.

Lookup tables can also be kind of slow, have you considered alternatives to the lookup? If they are simple (2-d, maybe 3-d) you may be able to fit an equation to the data and use that instead. Might be faster.

I do want to reiterate what I mentioned earlier as this 1000 input line throws a red flag in my mind. Although it is fully possible this is the best/fastest way to do the operation you desire, whenever I see the same thing being repeated, I have to step back and look at the problem to make sure it's the best way to do it.