First things first (as napolean dynamite) - luckeee....

Ok, now the beef. I would not look at "threading" anything. You have 28 processors, so if you can split things up, utilize AIX's built in ability to distribute these processes among the 28 cpus. In a nutshell, split the data up into 28 parts, run the scripts 28 times on the 28 chunks of data, and join the results back.

If you are considering "parallelizing" whatever it is you are doing, then I highly recommend you start over and use MPI. If it must be in perl, look at parallel mpi interface for Perl.

If you describe what you are doing (and want to write a parallel app), I am sure some of us can suggest a domain/task decomposition scheme would help you out.

BTW, bioinformatics screams, "parallelize me!!" :)