pipelined parallel enable table function

 Hi All,
For 'pipelined parallel enable' table function, do I have to enable session parallel query first?

I have read some white papers or web pages on Map and reduce implemented with Oracle table function, and find that in table function,

we have to use cursor in the loop to get one row and pipe it out. This changes SQL to pl/sql.

is it the cost that must be paid in table function?

Last, how can I confirm that Oracle has parallelized the table function?

Best regards,

Leon