Distinct count measure - 2000 Vs 2005

Hi,
When we implement distinct count measure in 2005 it is giving performance issues.
1. How to improve the performance of distinct count measures?
2. How to design aggregation for distinct count measure?

This probably won't answer your question directly, but have you update the statistics on the 2005 instance after restoring it?
I would start with defragging the indexes that need it, and updating the stats once I brought it into 2005. I don't know how you are getting it there though.

When is the performance is it during Reporting / Querying (Is end user observing the performance problems during analytices) or during processing of cube.
Since you are talking about Distinct Count assuming that there is a measure that is non additive and Distinct Count is the only option. In such cases it would be better if Distinct count measure is moved to its own measure group rather than clubbing with other measures. But dimensionality needs to be same for you analytical purpose.
May be if you can post more detials someone may be able to help you