------- Comment From [email protected] 2026-06-03 10:27 EDT------- Sven Schnelle bisected this to 7dadeaa6e851 ("sched: Further restrict the preemption modes").
Reverting that commit and going back to PREEMPT_NONE instead of PREEMPT_LAZY restores previous behaviour. Why PREEMPT_LAZY causes such a regression, and only with many CPUs is not yet understood. In theory it should hardly make any difference, but the numbers say it does. Using upstream vanilla kernel v7.1-rc6 (defconfig) with the above mentioned commit reverted gives the following results: Kernel compile (make -j $(nproc)) with PREEMPT_LAZY: real 7m32.483s user 181m54.293s sys 454m47.525s Kernel compile (make -j $(nproc)) with PREEMPT_NONE: real 1m58.805s user 104m56.404s sys 32m1.667s Further analyis is required, but at least a possible temporary fix is known. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2154748 Title: [Ubuntu 26.04] Severe Performance Degradation on kernel 7.0.0-15 To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu-z-systems/+bug/2154748/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
