Reduced, and can cause heavy traffic on the LLC cache. Other hand, the complexity of per-cpu updating at the tick is not The complexity is reduced by O(n) in the domain scope. The periodic load balancing has the admission control that only theįirst idle cpu (if any) or the balance cpu can trigger balance, so Now we update the filter during load balancing at SMT level, while it wasĭone at the tick (patch v1-v3). Sched-feature is under rework or been replaced (by SIS_UTIL ). The filter should be put under SIS_PROP which is not now, because this ![]() More overloaded the system is, the less cpus will be scanned. The filter only contains the unoccupied cpus,Īnd is updated during SMT level load balancing. This patchset introduces the SIS filter to help improving scan efficiency The number of cpus to be scanned, but the way of how it scans is not SIS_PROP was born to limit the scan cost. But thisĭoesn’t scale well once the LLC gets bigger and the load increases, so The cpus of the LLC due to lots of idle cpus can be available. The domain scan works well under light workload by simply traversing LLC (domain scan) if the hot ones are not idle. The cache hot cpus will be checked first, then other cpus of that The SIS tries to find an idle cpu inside that LLC to place the woken-up In maximizing the usage of cpu resources and can greatly affect overall ![]() The wakeup fastpath (select_idle_sibling or SIS) plays an important role To: Peter Zijlstra, Mel Gorman, Vincent GuittotĬc: Josh Don, Chen Yu, Tim Chen, K Prateek Nayak, Gautham R. ` (8 more replies) 0 siblings, 9 replies 47+ messages in thread 12:04 ` sched/fair: default to false in test_idle_cores Abel Wu Sched/fair: improve scan efficiency of SIS All of help / color / mirror / Atom feed * sched/fair: improve scan efficiency of SIS 12:04 Abel Wu
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |