Perl can't automatically make execution of builtins run parallel on multiple processors.

So your best bet is to use threads to parallelize it manually. But beware that it might not very easy, depending on what your program actually does.

But it makes me wonder - why does it work fast on the linux boxes? I'm pretty sure it works on just one CPU there as well.

Or maybe your bottleneck is somewhere else, like slow external resources? Try to profile your program (and look at the wallclock seconds, not just at the CPU seconds).