Francesco Casella

OK, it seems that other tool vendors than OSMC are fine with the current status. At the end of the day ModelicaServices.inf is defined as "Maximum representable finite floating-point number"...

Thanks @perost for the analysis! > The majority of the time is spent in `NFComponentRef.toString` I don't know the details of what is done in SimCode, but to me a...

@phannebohm I made some further reflections. It is probably worth making this hash thing more efficient. However, I understand the important scale factor here is the number of states (->...

I went back to this ticket and thought a bit more about it. @perost wrote: > Both of my PRs are now merged, and I can't really see any other...

I just checked the generated code and I think I have some clues. When comparing the code generated for the 3200 case and the 6400 case, most C source code...

#12914 may also have a positive effect on this issue, particularly once we've disabled the linearized model generation in large models.

@mwetter these improvements are probably also relevant for you @ LBL

After the fix in #12981, both backend time and simcode time were slightly reduced. Also the C compile time was reduced. Nothing dramatic. Before #12981: | Model | Backend time...

@phannebohm please see my [previous comment](https://github.com/OpenModelica/OpenModelica/issues/12627#issuecomment-2366970537) again. I understand SimCode (and also Templates) should be vectorized and run in a flash. Instead the time to run it sill grows as...

Maybe it's just handling the info.json and init.xml files? Those grow as O(N) and end up being disproportionally large, e.g. 60 and 211 MB for CascadedFirstOrder_N_204800