Simulation difference between r2007b and r2012b

Hello,

i recently received a new licence for r2012b and i am using simulink models that were developped and used with r2007b.

In particular i have a problem with a small function (which calculates a sliding average when activated and with an external sample number) that we developped using an 'Enable Subsystem' and 'SwitchCases' and Switchcase Action subsystems --> Even if the parameters and models are exactly the same, the outputs of r2007b and r2012b are definitely not the same.

My questions: - is there any known problem due to import in r2012b of r2007b models? - were there any changes in the definition of these blocks and their parameters?

Thanks!

Matthieu

[Kaustubha Govind]

Matthieu: It might help if you explain how they're different. Is the different of the order of 1e-5 for instance, or significantly different?

Answer 1

[Andreas Goser]

I have worked on a couple of occasions with the similar symptoms. It is very diffcult to resolve without having the model, whoch is why I recommend contacting Technical Support. This is what root causes I have found in different cases:

• Difference was caused by different computer architectures (32/64 bit) and there a good numerical reasons why this can happen
• Difference was caused by different c compilers in cases where e.g Simulink Accelerator was used
• Difference was caused by different processor manufacturers and their respective BLAS library
• Difference was caused by a change of a default option of a MATLAB command that was called

Answer 2

[Matthieu]

Thank you for your comment/message.

I just solve the problem (finally!), much easier than what you actually describe Andreas: the parameters of one output port has changed, don't know if it was because of the version change or an unfortunate "set_param" during a previous update... Instead of being reset to 0, the ouput value of an enabled subsystem is held. Hard to find when you have 250 such subsystems! One day lost for such a stupid thing, aarrrrgggg.

Sorry... and thank you for your answers!