Save total computing time into oo_

What's the logic for this? It's not part of the model, rather a result. Moreover, I am not sure we should reproduce results from the log-file. When would this be useful?

For instance, currently I am comparing the performance of different optimizers and samplers in the estimation of a model and would like to quickly assess computing time. So I am running say 20 different estimations of the same model, each with its own mod file on a Cluster. When everything is finished, I then want to create a table with the total computing times. Accesssing this from the _results.mat is easier than what I currently have to scrape it from the log file. So that is my use-case.

It is part of the mod file, so I think putting it into M_ is reasonable.

changed milestone to %5.x

Usually, we store in M_ things that are known at run time, while oo_ stores stuff that is determined when running the model. Also, M_ should be invariant to e.g. machines or Matlab/Octave versions where the code was executed. So I would prefer it in oo_.

added 1 commit

• c79ae5d4 - Save 'Total computing time' into oo_.time

Compare with previous version

changed title from Save total computing time into M_ to Save total computing time into oo_

Okay, I see your point and force pushed a new commit.

mentioned in commit ce09f662

merged