Not sure if this thread is still going but I find myself in a similar situation.
Server maxdop = 0, It's a small 8 core box with 16gb ram
I've got a 30gb table, 100columns wide, contains random indexes about 29, a couple of FK's , clustered index , 0 fragmentation as I just rebuilt all the indexes, the table has a trigger and a schema bound view referencing it and a computed column.
If I create or rebuild an index specifying maxdop 2,6,8 or 0. It doesn't use a parallel plan.
If I create an exact copy of the table, just a different name. Same FK, Views, Triggers,Computed column,Filegroup etc.
I run the same code with the maxdop 0,2,4,6 etc. I get the corresponding parallel threads every time.
So basically for two identical tables aside from their name and the fact that I created one new as opposed to a legacy old table with FK's turned on and off etc. It's been around the block.
I get different parallel behavior using the same code, not this is all within the same database. Behavior persist after reboots.