News for hackers. Stuff that matters.
Re:Runtime Support for Multicore Haskell: A Retrospective by zzzcpan on Tuesday December 17, 2019 @12:47PM (#21816249)
No, work stealing is pretty high overhead.
Re:Runtime Support for Multicore Haskell: A Retrospective by chrisseaton on Tuesday December 17, 2019 @12:53PM (#21816329)
That's not my experience. I think work stealing is low overhead as it amortises synchronisation, only does it at all when actively required to get more work, and reduces conflicts due to two ends of the queue. Why do you think it's high overhead?
Re:Runtime Support for Multicore Haskell: A Retrospective by zzzcpan on Tuesday December 17, 2019 @01:02PM (#21816454)
It's only one part of the story (scheduler) and not even good enough idea considering non-locality overhead.
Re:Runtime Support for Multicore Haskell: A Retrospective by chrisseaton on Tuesday December 17, 2019 @03:26PM (#21818397)
Isn't work stealing excellent for locality? Jobs stay on the same core until there's a need to steal, and then the most likely non-resident jobs are taken. Is it even cache oblivious?
Re:Runtime Support for Multicore Haskell: A Retrospective by BubRoss on Tuesday December 17, 2019 @03:07PM (#21818159)
Why would it be high overhead if work is only stolen when a core has run out of work?