When using a live connection to an SSAS tabular model, what sort of performance difference (if any) would there be between creating a report level measure and using a measure created within the tabular model itself? This is assuming that the DAX is identical between the two, of course.
I've created a report level measure using ALL to remove a particular filter in a drillthrough, and the performance is pretty terrible compared to any other measure in the model. I'm just trying to determine whether this would be improved if I created the same measure in the model. I would test this myself, but this particular model takes about 4 hours to process and must be completed by first doing a Process Clear due to memory limitations on the VM, and we have multiple people actively using this cube, so I can't bring it down at the moment.
