Is there a computing notion of "thread explosion" where you start more threads to use idle processor capacity because existing threads are in wait state but then they all unpredictably go active and the pre-emption logic is bad so everything crashes?
Conversation
Replying to
I have in mind something that would be like "pre-emption hierarchy illegibility" where you can only impose a weak partial order on the priorities of a number of threads, all of which have execution constraints (deadlines, varying memory load, etc)
2
Replying to
This sounds like "thrashing". Not perfectly analogous but near. en.wikipedia.org/wiki/Thrashing
1
1
Replying to
Yeah I thought of that too. Thrashing might be a special case of what I'm getting at.
1
Replying to
This is maybe adjacent to "priority inversion" in scheduling systems.
en.wikipedia.org/wiki/Priority_
The wiki example is one way, it can also happen if there are a lot of L threads and the overhead of dealing with them prevents H threads from being dispatched.
Replying to
Do you mean something like:
en.wikipedia.org/wiki/Thunderin
?
At the lower (micro-architecture) level there is something called Replay Tornadoes
jilp.org/vol7/v7paper2.



