buildRapid​Accelerato​rTarget gets stuck during execution in matlab 2020a, but not matlab 2017

I am trying to upgrade a tuning script from Matlab 2017a to Matlab 2020a and I run into an error while running:

load_system('My_Model') %Silently load model 
sltest.harness.open('My_Model','My_Model_Harness');
>>RAE = Simulink.BlockDiagram.buildRapidAcceleratorTarget('My_Model_Harness');

the rapid accelerator build will progress normally until I get the message:

### Successfully updated the model reference simulation target for: My_Model

after this, the rapid accelerator build gets hung and won't exit unless I abort with ctrl+c.

Any idea from anyone what could potentially be causing the issue? Or some way of how to debug? This works fine on R2017a, so I imagine something changed with simulink. My model was created with R2017a, and I am making a copy with R2020a before running it.

Answer 1

[Matthew Rademacher]

My model had a very very large lookup table ( a 3d motor map ) as an input parameter. The solution is to basically NOT assign that variable as an FMU parameter, which compiles it as part of the binary (.dll/.so) instead of defining inputs for every single index of the massive table. So basically if anyone reading has this issue, check the sizes of your input parameters in the fmu build GUI.

Answer 2

[Pat Canny]

I recommend contacting MathWorks Technical Support. This question requires more information.

[Matthew Rademacher]

Actually, I know the answer now!

My model had a very very large lookup table ( a 3d motor map ) as an input parameter. The solution is to basically NOT assign that variable as an FMU parameter, which compiles it as part of the binary (.dll/.so) instead of defining inputs for every single index of the massive table. So basically if anyone reading has this issue, check the sizes of your input parameters in the fmu build GUI. This answer is thanks to @Yashodhar Veeralla , @Jon Lobo, and the folks at the Novi office providing excellent technical support!