You can either do inserts via PARALLEL DML or DIRECT load via the APPEND HINT. But not both. THere is more info in the Concept section of the manual.

I moved a schema into a separate database (around 15 Gig) in around 1 hour using the APPEND hint. As we only have 3 CPUs and the databases were on the same machine there was no real time benefit running with the PARALLEL hint.

You might want to do some benchmarking on your system to see which method will be faster.

Have Fun