(I expect for my $value (@array) to be the fastest, because it does the least unnecessary extra operations, but that's really just guessing)

#!/usr/bin/perl
use warnings;
use strict;
use Benchmark qw(:all);

my @bigarray;
for(1..100000) {
  push @bigarray, "A randomish string for testing arrays with realisti
+c data" ;
}

timethese(100, {
    while_shift => sub{
                      my @array = @bigarray;
                      while (my $value=shift @array){}
                   },
    while_pop   => sub{
                      my @array = @bigarray;
                      while (my $value=pop @array){}
                   },
    for_each    => sub{
                      my @array = @bigarray;
                      foreach my $value (@array){}
                   },
    for_loop    => sub{
                      my @array = @bigarray;
                      for (my $i=0;$i<=@array;$i++) {
                      my $value=$array[$i];
                    }},
});
[download]

$ perl a.pl
Benchmark: timing 100 iterations of for_each, for_loop, while_pop, whi
+le_shift...
  for_each:  4 wallclock secs ( 3.85 usr +  0.01 sys =  3.86 CPU) @ 25
+.91/s (n=100)
  for_loop:  7 wallclock secs ( 7.36 usr +  0.01 sys =  7.37 CPU) @ 13
+.57/s (n=100)
 while_pop:  5 wallclock secs ( 5.35 usr +  0.00 sys =  5.35 CPU) @ 18
+.69/s (n=100)
while_shift:  5 wallclock secs ( 5.38 usr +  0.00 sys =  5.38 CPU) @ 1
+8.59/s (n=100)
[download]

I think it's also true that whatever difference there might be among those alternatives, it will be impressively insignificant -- it will count for nearly nothing -- when seen in the context of any real-life script that actually does something useful within the while or for loop.