Hello, on HP-UX

SMON: Restarting fast_start parallel rollback

I shut a DB down badly by killing the DB service during a very large transaction. It's been up again for 2hrs but it's still recovering and nothing else will run with any reasonable speed. Lots of IO.

Is there any way of bypassing/shortcutting/stopping the recovery process? I am happy to abandon all work done since the last committed transaction pre-abort.

Also the UNDO tablespace is not releasing the space used by the uncommited blocks. How can I free-up the UNDO tablespace?