Your results can vary greatly depending on your data. Check out this code that gives you a pretty evenly distributed sample.

use Benchmark;

# Get 200 random 4-digit integers
my @array = (grep {/\d{4}/} map { 10000 * (sprintf "%0.4f", rand) } (1
+..400))[1..200];
print "N = " . @array . "\n\n"; # Be sure we have the right sample siz
+e

timethese(0, {
     sorty => \&sorty,
     space => \&spaceship,
     comp  => \&comp
});

sub sorty {
  sort @array
}

sub comp {
  sort {$a cmp $b} @array
}

sub spaceship {
  sort {$a <=> $b} @array
}
[download]

PRINTS:

N = 200

Benchmark: running comp, sorty, space, each for at least 3 CPU seconds
+...
comp:  3 wallclock secs ( 3.15 usr +  0.01 sys =  3.16 CPU) @ 1064618.
+52/s (n=3368453)
sorty:  3 wallclock secs ( 3.11 usr +  0.00 sys =  3.11 CPU) @ 1096416
+.69/s (n=3415338)
space:  3 wallclock secs ( 3.11 usr +  0.00 sys =  3.11 CPU) @ 1064940
+.23/s (n=3314094)
[download]

Looks like a pretty even race to me!