No ,we can't do partition on a cube with data...
The data in the cube A has to copy to other cube B.Then we
have to do partition on cube A..The queries on cube A has
to copy to cube B..By using a T-code RSZC..It improves the
performance by partition..