Does the performance improve if you change the query to use proper JOINs instead?

insert in table T_3_XYZ ( t_3_id , t_3_data_1 , t_3_data_2 )
  select distinct t_1_id , t_1_data_1 , t_2_data_2
    from T_1_DAT
    inner join T_2_KEY on t_1_id = t_2_id
[download]

Ideally the query optimizer would recognize that the original WHERE clause is identical to the INNER JOIN clause, but maybe it doesn't?