What is the fastest way of writing large Simulink output data to csv file

I've run 1000 simulations with very small time steps in Simulink and stored the data in Simulink.SimulationOutput. I'd like to extract some info from this output and write it to a CSV file for each case. I tried creating a table for each simulation in a for loop and used the following command to save 1000 CSV files:

filename = fullfile('Results', filesep, sprintf('out_%d.csv',j));
writetable(T, fname)

Apparently for some simulations I have uneven numbers of rows so Table didn't work for me. I don't know which cases have uneven numbers, and I ran my simulation on a cluster. So to inspect that I then I tried saving the output with the following command with an intention to carry it to my computer to find out which simulation result(s) causes the problem.

save('results', 'out', '-v7.3')

The problem is that this file is ending up to be much larger than I expected (it's still running and the last time I checked it was going over 15 GB). Even after saving this, I won't be able to open it on my PC due to the file size. And I'd like to process the data with another software, so I'd like to save each simulation as a separate CSV file. What is the fastest way to do this for a handful of variables with different sizes?

PS: I don't want to decrease my simulation sample time any further, so I have to work with what I have right now.

[Jan]

The CSV file will not be remarkably smaller. It is expected to be larger, because the v7.3 format compresses the data. In addition text files are useful only, if they should be read or edited by human. In all other cases they are a drawback, because they take longer to read and the conversion from the binary values to the decimal strings reduces the precision and takes time.

I suggest to save the results in pieces to a set of MAT files. I do not post this as answer, because it is not clear if this is applicable to your problem.

[swenia]

Jan, thank you for your comment. Per your suggestion, I tried saving each case as a separate MAT file and that was really fast! But the problem is, I still need it as a txt or a csv format because the software I need to process this data with does not read MAT files.

[Jan]

Does it accept another type of binary file formats?