lhoward wrote Shift, Pop, Unshift and Push with Impunity!, which gives you a shallow overview. There also is Perlguts Illustrated, which tells you about the Perl data types and how they are implemented at the C level.

Whenever it comes to optimizing Perl, the most time is gained by making the program do less work. This means, use an optimal algorithm instead of trying to micro-optimize how you use arrays or hashes.

If you want to look in the direction of micro-optimizations, first do a benchmark with Devel::Dprof before trying to guess where your program spends most of its time.

To backup what Corion said about optimizing and profiling, there's a good article on perl.com about profiling. Includes a decent real world example of taking a script's execution time from 10 seconds to 4 seconds.

Cheers,

Brent


-- Yeah, I'm a Delt.

As I commented in To push and pop or not to push and pop?, there is almost no difference in the performance of shift, unshift, push and pop. This is not a place to look for efficiency gains - writers of the Perl code have already done it :).

A more efficient approach is to profile. Take a look at some of the answers in Using Devel::Profile output to know what to do next to get some ideas (found using Super Search for nodes whose titles contain "profile" in SoPW).