Recently received the following question from a customer:
When running a scenario with the output set to “None” or “File” the CSV log varies in length and is often inaccurate compared to the set scenario time, why is this?
This is because with the output set to “None” or “File” Skydel will run the scenario as fast as possible and the csv logging sometimes can’t keep up with 100% accuracy. If an accurate CSV log is desired, it’s suggested to run the scenario with “NoneRT” selected as the output as Skydel will run this scenario in real time (RT) and the csv logging can keep up much more accurately.
Note: There may still be some slight variance in the CSV log time beyond the end of the scenario, but the log will be complete and accurate up to the completion time of the run.
The Skydel engine will not stop at the exact millisecond when the scenario reach the desired duration. Internally, Skydel look at the elapsed time from time to time and when the elapsed time is equal or longer than the scenario duration, it will stop. This action is asynchronous. Therefore, the log can be equal or slightly longer, but never shorter than the scenario duration time.
I just wanted to point out that this behaviour is the same for all output types. The difference with “None” and “File” outputs, is that they go as fast as possible. They are therefore able to execute more milliseconds of simulation during the asynchronous stop, which amplifies the log length difference.