Forrtl: severe (38): error during write, unit -1 file CONOUT$

I created the model in 2020. It was compiled and ran successfully. I was able to compile this model successfully with MATLAB 2024b and 2025a but error occured during runtime.

[Walter Roberson]

By the way, the warning is suggesting that you need to use g10.3 instead of g9.3

[Walter Roberson]

severe (38) error appears to be a symptom instead of the cause of the error. I am finding several potential causes for it, all completely different -- indicating that something else is going on in the appropriate code leading to internal problems that just happen to trigger an error 38.

For example I find one reference saying that for a particullar program, the problem is simultaneous action to temporary files. I find another reference saying that for a different program, the problem is differences between the system locale and the user locale. Another site suggests maybe you ran out of disk space. The same site suggests adding FLUSH_FILE_BUFFERS=.FALSE (which solved that problem for that particular user.)

[Tuan]

There is 637GB free. Where should I add FLUSH_FILE_BUFFERS=.FALSE ?

[Walter Roberson]

I am finding hints that FLUSH_FILE_BUFFERS might only apply to Fire Dynamics Simulation (FDS) programs.

Answer 1

[Tuan]

The problem is solved.

Google search indicated:This runtime error occurs when an Intel Fortran application fails to write output to the console or a standard output device. Error 38 specifically indicates a "write error" during an I/O operation. This suggested statements like Write(*, or Print*, in Fortran codes caused the problem. And I found WRITE(*,193)t,Thet ... in the codes that caused simulink crashed. Matlab 2024 had no problem with WRITE(*.