then there can be only one plausible explanation

Utter bolix. There can be many many reasons for bad performance and there is nothing like enough information in the OP's post to make the assertion that "your system is thrashing".

To clarify a little: I have tab-delimited files that usually end up being around half a million lines long each. Originally, I had set my code up so that it would create a new object for each line, where its "fields" were the attributes. This took "half a day", or to be less colloquial about 4 hours. Now, I'm just parsing each line as a hash of attributes into a set of hashes that stores all the lines have their type attributes in common. This is now taking 16 seconds. I don't think my system is thrashing. I just haven't quite learnt how to use Perl effectively!

I think there must have been some major (perhaps hidden) work going on in the OO version that you didn't notice. All else being equal, I doubt very much that adding OO wrappers in the fashion you describe would even add a few percent to the run time of your script - the I/O time alone would bury any OO overhead.

If you still have the OO version of the code (in your revision control system for example - what? no revision control? I'll wait while you install Mercurial of git) can you post a sanitised version of the code the exhibits extreme slowness along with a trivial amount of data? I'm interested to see if I can spot the time sink. Were you using an OO library like Moose perchance?

Aside from that it sounds like your current approach is appropriate. Again, as a learning exercise, you may like to post a sanitised version of your code for comment.

True laziness is hard work