Just in case this is useful - once upon a time I benchmarked lookup tables. The test looked up every table entry once in random order, then repeated a number of times to get an average. Here's a summary from my notes:

Comparison of lookup tables
Range: 2 to 128 byte strings, 10 to 100,000 values

1) By hash key lookup.

2) By iteration over an unsorted array with foreach, 
   exiting at first match.

3) By search algorithms on a sorted array.

Results:

1) For small string sizes, and larger strings with
   fewer than 1000 table values a hash key lookup 
   ranged from 0 to 33% faster than the unsorted array.

2) With 8 byte or longer strings the speed crossover
   between hash and unsorted array usually occurred
   with tables between 1000 and 2000 values. At 2000
   values the unsorted array lookup speed was:
    * 20% faster for 8 character strings
    * 50% faster for 32 character strings
    * 100% faster for 128 character strings

3) In all cases the sorted array was significantly
   slower regardless of search algorithm. Worst case
   was 4% of hash lookup speed, best case 17% of hash
   lookup speed.
[download]

This was done in ActivePerl 5.6 on a 1GHz 'doz 98 machine. Moral: in most cases you may be better off using whatever format your data is already in rather than converting between arrays and hashes.