I think I've said it here before but I'll say it again. There's fast and there's fast enough. Knowing the difference is what makes the difference between finishing early and keeping it simple and wasting time optimising something that will change once your line manager finally gets their requirements sorted.

Flippancy aside, if you're having to re-write massive chunks in inline::C then surely you should just treat your Perl app as the prototype and build the production app in C or C++.

Of course you've probably gone and then doubled the cost of your project but at least it runs faster now :)