PostgreSQL 10.3 coalesce vs. greatest

gistfile1.txt

drop table t;
create table t(c1 int, c2 int);
insert into t values (1, null), (2, null), (null, 7), (null, 4), (3, 20);

# select * from t;
 c1 | c2
----+----
  1 |
  2 |
    |  7
    |  4
  3 | 20
(5 rows)

# select max(coalesce(c1,c2)) from t;
 max
-----
   7

# select max(greatest(c1, c2)) from t;
 max
-----
  20