extensive waits on 'direct path writes'

From a statspack report (9.2.0.8 on solaris) of a 5 hour duration during a night batch update cycle on a data warehouse

Top 5 Timed Events

~~~~~~~~~~~~~~~~~~                                                     % Total
Event                                               Waits    Time (s) Ela Time
-------------------------------------------- ------------ ----------- --------
PX Deq Credit: send blkd                        1,717,992     121,654    17.92
direct path write                                 212,362     109,472    16.13
db file scattered read                          2,080,746     106,406    15.68
direct path read                                1,420,179      88,819    13.08
enqueue                                            28,024      79,127    11.66

I've never seen the 'direct path write' event be the top event before
(according to Metalink the PX Deq... event is an idle event, so I'm
ignoring that one)

I'm trying to understand the following statements (from metalink) "If asynchronous IO is not being used then the IO write request blocks until completed but this dies not show as a wait at the time the IO is issued. The session returns later to pick up the completed IO data but can then show a wait on "direct path write" even though this wait will return immediately. "

1. How do I know for sure if async IO is being used
   (I didn't set up the system, I'm just being asked to trouble shoot)
   
2. Since the statement above seems to say that if async IO is NOT being used, then we shouldn't see waits for this event, does that mean that we definately are using async IO?
3. Does the phrase "this wait will return immediately" imply that no time will accumulate for this event when async IO is not being used? Any enlightenment gratefully accepted.

Matt

--
http://www.freelists.org/webpage/oracle-l