As per the Application optimization technique Count( Distinct ‘FieldName’) will have impact in performance. It will have the impact even if i use this in the conditional show of the object.
I haven't seen a performance issue using a "count distinct" in a while. Granted, I'm not using massive data sets, but I am thinking I read somewhere once where this isn't as big of an issue as it used to be.
Anyway, you're correct ~ if you're conditionally showing/hiding an object it is being calculated even though you don't see it. For some objects (e.g. charts) this can be worked around by using the same show/hide condition as the Calculation Condition in the object's properties (I basically do this will all of my conditionally shown charts to maintain performance.)
Alternatively; "back in the day" I would use a counter field in a dimension table (where each record has a value of "1" in the counter field) to increase performance vs. count distinct, and sum() on said field instead of count().
Thanks for your response Isaiah Weed, It is there in the performance document. As the function will run on only one cpu even the server multiple cpu cores. Mine is very large dataset. I couldnt find the way to test it out.
I have taken the mem statistics. But it is not having the details of the CPU cores only cpu core is used or all the cpu core is used. ideally i want to understand the QVW file using the multithread rather only one thread is used for calculation it will impact the performance.
