It was rather difficult to determine what issue you were commenting about. You indicated:
We ran a PB Trace in PB 12.5.1 for a first run for a specific criteria and then again for the second run. All that it shows me is that the EXECUTE and each FETCH for the report in PB 12.5.1 when it was run for the first time took considerably longer than the second time we ran the report fro the same criteria in the PB 12.5.1 version.
And I explained why that would happen.
If you want to see why there is a difference between 9 and 12.5 after both statements have run more than once, the first thing I would look at is the SQL statement that actually got generated. Perhaps there is some difference there.
Otherwise, as Chris indicated, the issue is likely to be more data (resulting from Unicode versus ANSI) and potentially more overhead that 12.5 uses versus 9.0.