我有分区表，表结构

create table tab1 
(
col1 int,
col2 string,
...
col50 int,
col51 int
)
partitioned by 
(col50 int, col51 int)
stored as orc;

目前我们有约17000个分区，每个分区至少有约50k条记录。
下面的查询需要更多时间~90分钟

SELECT DISTINCT col2 FROM tab1
select col2 from (select col2, row_number() over (partition by col2 order by col3) as rnk from tab1) t1 where t1.rnk=1

有没有办法缩短执行时间，提前谢谢