hi, Peter Zijlstra, we know from Yu Chen that this commit could benefit other workloads like blogbench/hackbench. since by more runs of this stress-ng test, we still observed stable small regression. we just report this result FYI. Hello, kernel test robot noticed a -3.1% regression of stress-ng.dirdeep.ops_per_sec on: commit: 63304558ba5dcaaff9e052ee43cfdcc7f9c29e85 ("sched/eevdf: Curb wakeup-preemption") https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master testcase: stress-ng test machine: 36 threads 1 sockets Intel(R) Core(TM) i9-9980XE CPU @ 3.00GHz (Skylake) with 32G memory parameters: nr_threads: 10% disk: 1SSD testtime: 60s fs: xfs class: filesystem test: dirdeep cpufreq_governor: performance If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot | Closes: https://lore.kernel.org/oe-lkp/202309041403.9a7531ba-oliver.sang@intel.com Details are as below: --------------------------------------------------------------------------------------------------> The kernel config and materials to reproduce are available at: https://download.01.org/0day-ci/archive/20230904/202309041403.9a7531ba-oliver.sang@intel.com ========================================================================================= class/compiler/cpufreq_governor/disk/fs/kconfig/nr_threads/rootfs/tbox_group/test/testcase/testtime: filesystem/gcc-12/performance/1SSD/xfs/x86_64-rhel-8.3/10%/debian-11.1-x86_64-20220510.cgz/lkp-skl-d08/dirdeep/stress-ng/60s commit: 7170509cad ("sched: Simplify sched_core_cpu_{starting,deactivate}()") 63304558ba ("sched/eevdf: Curb wakeup-preemption") 7170509cadbb76e5 63304558ba5dcaaff9e052ee43c ---------------- --------------------------- %stddev %change %stddev \ | \ 7.85 -2.9% 7.62 iostat.cpu.system 0.14 ± 20% +0.1 0.22 ± 21% perf-profile.children.cycles-pp.pick_next_task_fair 12528411 +100.0% 25056823 sched_debug.sysctl_sched.sysctl_sched_features 2874027 -3.1% 2785140 stress-ng.dirdeep.ops 47338 -3.1% 45876 stress-ng.dirdeep.ops_per_sec 16652 ± 3% -10.6% 14890 ± 4% stress-ng.time.involuntary_context_switches 269.05 -3.0% 260.90 stress-ng.time.percent_of_cpu_this_job_got 164.73 -3.0% 159.74 stress-ng.time.system_time 0.01 ± 19% +1659.6% 0.20 ± 6% perf-sched.sch_delay.avg.ms.worker_thread.kthread.ret_from_fork.ret_from_fork_asm 0.00 ± 20% -42.0% 0.00 ± 34% perf-sched.sch_delay.avg.ms.xlog_force_lsn.xfs_log_force_seq.xfs_file_fsync.__x64_sys_fsync 0.19 ± 4% +187.6% 0.55 ± 6% perf-sched.wait_and_delay.avg.ms.__cond_resched.dput.terminate_walk.path_openat.do_filp_open 0.21 ± 7% +129.8% 0.49 ± 7% perf-sched.wait_and_delay.avg.ms.schedule_timeout.__wait_for_common.__flush_workqueue.xlog_cil_push_now.isra 0.19 ± 10% +204.1% 0.59 ± 11% perf-sched.wait_time.avg.ms.__cond_resched.__kmem_cache_alloc_node.kmalloc_trace.xfs_vn_get_link.pick_link 0.19 ± 4% +189.1% 0.55 ± 6% perf-sched.wait_time.avg.ms.__cond_resched.dput.terminate_walk.path_openat.do_filp_open 0.21 ± 7% +131.1% 0.49 ± 7% perf-sched.wait_time.avg.ms.schedule_timeout.__wait_for_common.__flush_workqueue.xlog_cil_push_now.isra 0.28 ± 28% +235.2% 0.94 ± 20% perf-sched.wait_time.avg.ms.xlog_wait_on_iclog.xfs_file_fsync.__x64_sys_fsync.do_syscall_64 3.693e+09 -2.7% 3.593e+09 perf-stat.i.branch-instructions 1.174e+10 -2.4% 1.146e+10 perf-stat.i.cpu-cycles 5.923e+09 -2.8% 5.756e+09 perf-stat.i.dTLB-loads 3.172e+09 -2.9% 3.081e+09 perf-stat.i.dTLB-stores 2.104e+10 -2.8% 2.046e+10 perf-stat.i.instructions 0.33 -2.4% 0.32 perf-stat.i.metric.GHz 358.39 -2.8% 348.30 perf-stat.i.metric.M/sec 3.635e+09 -2.7% 3.536e+09 perf-stat.ps.branch-instructions 1.156e+10 -2.4% 1.128e+10 perf-stat.ps.cpu-cycles 5.831e+09 -2.8% 5.666e+09 perf-stat.ps.dTLB-loads 3.123e+09 -2.9% 3.033e+09 perf-stat.ps.dTLB-stores 2.071e+10 -2.8% 2.014e+10 perf-stat.ps.instructions 1.319e+12 -3.1% 1.279e+12 perf-stat.total.instructions Disclaimer: Results have been estimated based on internal Intel analysis and are provided for informational purposes only. Any difference in system hardware or software design or configuration may affect actual performance. -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki