While we're on the topic of benchmarking, I'd just like to, for the hell of it, add the worst possible answer to this question:

map { $code .= "\$hash{$_} = 1;\n" } @array; eval "$code";
[download]

And now for a little program, borrowed from httptech:

use Benchmark;

my @array = qw(a b c d e f g h i j k l m n o p q r s t u v w z y z);
my %hash;

timethese(10000, {
        'btrott' => sub {
                @hash{@array} = (1) x @array;
        },
        'plaid' => sub {
                my $code
        map { $code .= "\$hash{$_} = 1;\n" } @array;
        eval "$code";
        }
});
[download]

Outputs

Benchmark: timing 10000 iterations of btrott, plaid...
    btrott:  1 wallclock secs ( 0.56 usr +  0.00 sys =  0.56 CPU)
     plaid: 23 wallclock secs (22.07 usr +  0.06 sys = 22.13 CPU)
[download]

Don't know why i threw this in. Guess I just didn't want to waste the effort:)