想爆头的sql查询,sql性能优化

DECLARE @table TABLE
(
id INT PRIMARY KEY,
bundle NVARCHAR(20),
section NVARCHAR(20)
)
INSERT INTO @table VALUES(1,'001','A')
INSERT INTO @table VALUES(2,'001','A')
INSERT INTO @table VALUES(3,'002','A')
INSERT INTO @table VALUES(4,'002','B')
INSERT INTO @table VALUES(5,'003','B')
INSERT INTO @table VALUES(6,'003','A')
INSERT INTO @table VALUES(7,'003','C')
INSERT INTO @table VALUES(8,'004','A')
INSERT INTO @table VALUES(9,'005','D')
SELECT * from @table
示例表(id为主键)
id bundle section
1 001 A
2 001 A
3 002 A
4 002 B
5 003 B
6 003 A
7 003 C
8 004 A
9 005 D

问1:筛选出bundle相同,但section不同的行,
  只有一个bundle的不用查出来,请注意,
    该表数据量非常大,必须要考虑性能
    (可以提出索引优化等建议)
筛选结果应为:
id bundle section
3 002 A
4 002 B
5 003 B
6 003 A
7 003 C

---

SQL code:
SELECT * from tb AS A
WHERE EXISTS(SELECT * from tb
WHERE bundle = A.bundle AND section <> A.section