@alist=qw/1 50 2 658 74 256 21/;
($greatest) = sort {$b<=>$a}@alist;
print $greatest;
[download]

Will also work for sort {$b<=>$a} keys %myhash

addendum:

For extremely large sets, matija is absolutely right:

use Benchmark ;

my @a = (1..500000);
my $greatest;
timethese(5, {
      'sort' =>
      sub {($greatest)=sort{$b<=>$a}@a;},
      'grep' =>   
      sub { 
          grep($greatest=($_>$greatest)?$_:$greatest,@a); 
          },
});


Benchmark: timing 5 iterations of grep, sort...
      grep:  3 wallclock secs ( 3.36 usr +  0.02 sys =  3.38 CPU) @  1
+.48/s (n=5)
      sort: 13 wallclock secs (12.22 usr +  0.04 sys = 12.26 CPU) @  0
+.41/s (n=5)
[download]