One need have atleast a unique column such as timestamp col 
(and assumption is to keep lowest tmpstmp) OR some  key col 
say IPID (again keep lowest value)..

One determined - Have a nested Select on all rows (except 
that key col) with group by rest of the columns + having 
count(*) > 0 + aggreate MIN(key_col).

    Now said that, have another outer SEL on all columsn & 
do a inner join with above nested Sel .. WHERE outer 
key_col <> MIN value of nested SEL..

See if it works..

Nested query method might be required in other databases 
how ever in TD we don’t need to follow such a difficult way 
to just find out the unique rows. 

In TD we have functions like Rank () and Rownum() in the 
combination of Qualify, helps you to select out the rows  
which you wants to delete.
 
you can add a condition like ‘Where Rank() > 1’

Query to find duplicates in a table:(Custname, Prod, 
Order_amt)

select custname,count(*) from sales1 a where a.rowid > ANY 
(select b.rowid from sales1 b where a.custname=b.custname 
and a.prod=b.prod and a.order_amt=b.order_amt) group by 
custname;

Query to delete duplicates:

delete from sales1 a where a.rowid > ANY (select b.rowid 
from sales1 b where a.custname=b.custname and a.prod=b.prod 
and a.order_amt=b.order_amt);

guys rowid concept is discontinued in teradata as far as my
knowledge goes, we can always use below sql ...

INSERT INTO nodupes_table ( all_columns )
     SELECT all_columns
     FROM dupes_table
     QUALIFY ...

where the QUALIFY ... can be (depending on your version):
     /* V2R5 syntax */
     QUALIFY ROW_NUMBER() OVER (PARTITION BY all_columns
ORDER BY all_columns) = 1

     /* or V2R4 and higher equivalent functionality */
     GROUP BY all_columns
     QUALIFY CSUM(1, all_columns ) = 1

     /* or, alternative OLAP SUM V2R4 and higher syntax */
     QUALIFY SUM(1) OVER (PARTITION BY all_columns ORDER BY
all_columns ROWS
     UNBOUNDED PRECEDING ) = 1


I do think an insert-select into a set table would be a
cleaner process (don't know about runtime, though):
     INSERT INTO nodupes_set_table ( all_columns )
     SELECT all_columns
     FROM    dupes_table;


refer to teradata forum for more info

The following example shows how to create table and insert 
data without duplicate rows from existing table:

CREATE TABLE NODUPS AS
(
	SELECT * FROM Departments
	UNION
	SELECT * FROM Departments
)
WITH DATA;