The parallel execution model uses two sets of slave processes called producers and consumers. Some of the data that is passed from producers to consumers are intermediate values, rather than part of the final result. The PIV_SSF hint occurs in producer SQL returning intermediate values for a single set function like COUNT, SUM, MIN or MAX. The PIV_GB hint is used where similar intermediate values are being returned for a GROUP BY sort. CIV_GB is the corresponding hint in the consumer SQL. The SWAP_JOIN_INPUTS hint is found in slave SQL for hash joins. I don't know why it is needed, because hash join processing normally allows the swapping of the join inputs anyway.

When examining TKPROF output from queries that have parallelism, the SQL of the parallel slave processes is exposed. I have observed the following 4 undocumented hints in these SQL statements. Do you know what they are?

CIV_GB, PIV_GB, PIV_SSF, SWAP_JOIN_INPUTS