I’ve received a response from the MS Support Engineer about the issue in processing PC dimensions (see earlier post: Parent Child Dimension Processing Time).
Here it is:
"Development has been looking at the issue wherein dimensions having P/C hierarchies cause validation of the dimension structures of all partitions even when no changes have occurred, and they indicated that it is a pretty major design change to implement for SP2 or a QFE. Currently, they have this scheduled for the next release of SQL Server, due out sometime in 2007. Would this be acceptable to you? It doesn’t seem likely a QFE will be approved to have the issue fixed for SP2 of 2005 this late in the process, unfortunately, but they have agreed to fix it in the next version of the product."
How could they overlook something like this? We have a database with appx 5k partitions adding more daily and this bahavior adds four-five hours of processing a day. Processing a PC dimensions consists 80% of our daily processing time. It’s killing us.
So be careful on how you use PC dimensions in your SSAS2005 database. While a small scale enviroment should be fine, those of us with larger environments will want to avoid using PC dimensions if possible.
Jesse Orosz: SQL Server Analysis Services Blog 
You could try to use the Multiple Hierarchies scenario I described in my paper (http://www.sqlbi.eu/manytomany.aspx).
You can limit the dependencies on a single bridge measure group, that will be rebuilt each time.
We use a similar scenario in a production environment. Performance are similar, since a P/C dimension does not create aggregation on intermediate levels anyway.
Let me know [marco.russo At sqlbi(dot)eu] if you try and give me any feedback, it would be useful in a future version of my paper.
Marco Russo
http://www.sqlbi.eu
http://www.sqljunkies.com/weblog/sqlbi