The below query will give you some stats about the data distribution as how many rows will be landing on different amps... Execute the query with different column list, you surely will be able to choose the ones that are more appropriate...
SELECT HASHAMP(HASHBUCKET(HASHROW(<PI_COLUMN_LIST>))) AS "AMP#",COUNT(*)
FROM <TABLE_NAME>
GROUP BY 1
ORDER BY 2 DESC;
The below query will give you some stats about the data distribution as how many rows will be landing on different amps... Execute the query with different column list, you surely will be able to choose the ones that are more appropriate...