With this fight over spinlocks in Linux almost clear, now I want to know if @Jonathan_Blow used them in their games and if he knew about the performance hit
-
-
Because you have ensured that the _common case_ is the exact scenario that will cause weird things like getting swapped out in between a clock sample and a exchange, even though that is _never_ the common case in a well-written multithreaded program.
-
In order to time a scheduler, you have to have your threads _actually do something_, otherwise you are timing something that you should have designed your program to avoid, which is 100% of your threads contending the lock 100% of the time. Who cares what a scheduler does then?
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.