Yet Another Deleting Duplicate Records (in-place)

USE tempdb;
GO

CREATE TABLE MyTab (Id1 int, Id2 varchar(1));
GO

INSERT MyTab
SELECT 1, 'A' UNION ALL
SELECT 1, 'B' UNION ALL
SELECT 1, 'B' UNION ALL
SELECT 1, 'C' UNION ALL
SELECT 1, 'D' UNION ALL
SELECT 2, 'A' UNION ALL
SELECT 2, 'A' UNION ALL
SELECT 2, 'A' UNION ALL
SELECT 2, 'B' UNION ALL
SELECT 2, 'C'
GO

SELECT * FROM MyTab;
GO

ALTER TABLE MyTab
ADD rowguid uniqueidentifier NOT NULL
CONSTRAINT DF__MyTab__rowguid DEFAULT (NEWID());
GO

SELECT * FROM MyTab;
GO

DELETE t -- SELECT t.* 
FROM MyTab t -- your table name
JOIN (
 SELECT Id1, Id2 -- list of must-be-unique columns
 , MAX(CAST(rowguid AS varchar(100))) AS max_rowguid
 FROM MyTab -- your table name
 GROUP BY Id1, Id2 -- list of must-be-unique columns
 HAVING COUNT(*) > 1
) t1
 ON t.Id1=t1.Id1 -- add all must-be-unique columns 
 AND t.Id2=t1.Id2
WHERE t.rowguid <> t1.max_rowguid;
GO

ALTER TABLE MyTab DROP CONSTRAINT DF__MyTab__rowguid;
GO
ALTER TABLE MyTab DROP COLUMN rowguid;
GO

SELECT * FROM MyTab;
GO

DROP TABLE MyTab;
GO