#10454: New scheduler: substantial performance drop -----------------------------+---------------------------- Reporter: stippi | Owner: pdziepak Type: task | Status: in-progress Priority: high | Milestone: R1 Component: System/Kernel | Version: R1/Development Resolution: | Keywords: Blocked By: 10487 | Blocking: Has a Patch: 0 | Platform: All -----------------------------+---------------------------- Comment (by pdziepak): There was a delay before thread migration and effect of that migration becoming visible in the load statistics. That caused scheduler in some circumstances to migrate virtually all active threads to a single core (the best fit at that specific moment) then, after updating load statistics, the scheduler would realize that there is an overloaded core and migrate most of the active threads to a single core that currently is the best fit. Basically, this would cause each core to be overloaded for a short period of time while the others are idle. Average load would look like none of the cores reaches 100% load. The situation will eventually stabilize as load statistics updates were not synchronized in any way with thread migration and that's why the problem doesn't significantly affect long lived threads (e.g. that WonderBrush prototype). Anyway, hrev46824 should make things at least a bit better. -- Ticket URL: <https://dev.haiku-os.org/ticket/10454#comment:12> Haiku <https://dev.haiku-os.org> Haiku - the operating system.