So having a FILEGROUP to store the indexes and another for the data isn't an advantage?

It just useful for "organizing" the database?

But FILEGROUP for different types of data is useful as long as the indexes are in the same FILEGROUP then?

If the number of files for tempdb is related with the number of CORES why doesn't the rule apply for database files?

Is it because tempdb has "load" coming from "all directions" (all databases, DMVs, ...), hence having more IO?

Thanks,

Pedro