This is nice but it misses the case where the pair of numbers is equal, as the first test will pass, but the second will fail. I believe this small modification will cover that case though

use Data::Dump qw[ pp ];

my @v = ([100, 200],[200, 300],[200, 100], [200, 200]);;

my %uniq; 
my @r = grep{
    ++$uniq{ join $;, sort @$_ } == 1 
} @v;;

pp @r;;
#([100, 200], [200, 300], [200, 200])
[download]

- Miller