The solutions provided by igelkott both produce the same output for the example input data:

2 - 1 1 1 1 1
1 - 1 1 1 2 8
1 - 1 5 6 12 12
[download]

However, this output does not conform to the example output given by the OP:

2 - 1 1 1 1 1
1 - 1 2 1 1 8
1 - 1 6 5 12 12
[download]

The following code produces output that seems to conform to the OP's exemplified requirement (the two added values of '0 0 0 0 0' test for proper handling of leading-zero removal):

2 - 0 0 0 0 0
2 - 1 1 1 1 1
1 - 1 2 1 1 8
1 - 1 6 5 12 12
[download]

Code:

use warnings;
use strict;


my @AllPatterns = (
    '1 2 1 1 8',
    '0 0 0 0 0',
    '1 6 5 12 12',
    '0 0 0 0 0',
    '1 1 1 1 1',
    '1 1 1 1 1',
    );

{  # limit scope of array processing variables

my $biggest = 10;  # biggest number (most decimal digits)
my $fmt     = join ' ', ("%0${biggest}ld") x 5;
my $lead0s  = qr{ \b 0{1,@{[ $biggest - 1 ]}} }xms;

my %seen;  # unique-ifing hash

@AllPatterns =                     # 7. save to original array
    map  { "$seen{$_} - $_"      } # 6. make into final format
    grep { not $seen{$_}++       } # 5. only patterns not seen yet
    map  { s{$lead0s}{}xmsg;  $_ } # 4. remove padding
    sort                           # 3. lexicographic ascending sort
    # map  { print "'$_' \n"; $_ } # 2b. FOR DEBUG
    map  { sprintf $fmt, split   } # 2. pad fields to constant widths
    @AllPatterns                   # 1. for all patterns...
    ;

}  # end scope of array processing variables

print join("\n", @AllPatterns), "\n";
[download]