Not the way I tested it?

#! perl -slw
use strict;
use Hash::Util qw[ lock_hash unlock_hash ];
use Benchmark qw[ cmpthese ];

our $ITERS ||= -1;

our @items = qw[ age old years ];
our @tests = ( @items, map{ $_ . 'x' } @items );

our %set = map{ $_ => 1 } @items; lock_hash %set;
our $set = $;  . join( $;, @items ) . $;;

cmpthese $ITERS, {
    hash => q[
        my $count;
        exists $set{ $_ } and ++$count for @tests;
        print "hash: $count" if $ITERS == 1;
    ],
    index => q[
        my $count;
        1+index $set, $; . $_ . $; and ++$count for @tests;
        print "index: $count" if $ITERS == 1;
    ],
};

@items = map{ $_ x 5 } 'a' .. 'z';
@tests = ( @items, map{ $_ . 'x' } @items );

unlock_hash %set;
%set = map{ $_ => 1 } @items; lock_hash %set;
$set = $;  . join( $;, @items ) . $;;

cmpthese $ITERS, {
    hash => q[
        my $count;
        exists $set{ $_ } and ++$count for @tests;
        print "hash: $count" if $ITERS == 1;
    ],
    index => q[
        my $count;
        1+index $set, $; . $_ . $; and ++$count for @tests;
        print "index: $count" if $ITERS == 1;
    ],
};

__END__
## Check they produce teh same results:
C:\test>junk3 -ITERS=1
hash: 3
            (warning: too few iterations for a reliable count)
index: 3
            (warning: too few iterations for a reliable count)

                    Rate  hash index
hash  1000000000000000/s    --    0%
index 1000000000000000/s    0%    --
hash: 26
            (warning: too few iterations for a reliable count)
index: 26
            (warning: too few iterations for a reliable count)
                    Rate  hash index
hash  1000000000000000/s    --    0%
index 1000000000000000/s    0%    --


## See how fast
C:\test>junk3
          Rate index  hash
index 166820/s    --  -31%
hash  240174/s   44%    --
         Rate index  hash
index 18301/s    --  -59%
hash  44377/s  142%    --
[download]