Oh, stop. I see. Are you trying to track situation when thread chooses "wrong" thread for stealing? I.e. thread 1 tries to steal work from thread 2, but thread 2 also doesn't have any work. But there is thread 3 which do has some work to steal.
Hi Dmitriy
First off, thank you very much for your detailed responses. As for what exactly I am trying to track, yes, the false negatives are indeed the case when a thread chooses the "wrong" thread for stealing. That is precisely what I am attempting to track.
In terms of getting the number of spawned tasks, I implemented your suggestion and that definitely solves the occupancy problems I was seeing before. Thank you for mentioning the issue of mailboxes as well. I had completely overlooked this and it would be interesting to study this aspect too!
Abhishek
