If there is a desire to increase efficiency, I would get rid of the map which is actually a loop, and also the array indexing. Code shown below. Of course, as you suggest benchmarking is absolutely necessary if high performance is desired. The example test strings need to be much longer because now with just 5 characters, the setup code dwarfs the actual comparison code.

An XS procedure written in C could be done very efficiently. A goal would be to reduce the number of main memory cycles. I have a 64 bit machine. There is a bit of setup and cleanup code to focus on the 64 bit aligned block of memory. Read the buffers 8 bytes at a time. Do an XOR operation. If zero, all 8 bytes are the same. If not, then test each byte to see how many bytes differed. An assembly solution probably would provide significant performance increases over the C implementation. This is one of those cases where a human can probably easily beat the compiler. That's because there are some special buffer oriented instructions that are very difficult for the compiler to use effectively.

Anyway some code for comparison..

use strict;
use warnings;
use Data::Dumper;
use List::Util qw(min);


my @tests = (
    [qw{ABCGE ABCGE}],
    [qw{ABCGE FGCGB}],
    [qw{ABCGE JHAGT}],
);


foreach my $arry_ref (@tests)
{
    my ($str1,$str2) = @$arry_ref;
    
    # perhaps optional check to use shortest length
    my $len = min (length ($str1), length($str2));
    
    my $c_delta = 0;
    my $i =0;
    
    while ($i < $len)
    {
        $c_delta++ if (substr($str1,$i,1) ne substr($str2,$i,1));     
+  
        $i++;
    }
    
    print "$str1 $str2 $c_delta\n";
}
__END__
ABCGE ABCGE 0
ABCGE FGCGB 3
ABCGE JHAGT 4
[download]