edit: this should be one of the most basic queries ever. so, i have been surprised why you have choosen to add an identity column and querying the table twice with joins, etc.
shouldn't the script below suit your needs?
from t1 (NOLOCK)
group by y, x
having count(1) > 1