Greeting, FYI, we noticed a 2.3% improvement of reaim.jobs_per_min due to commit: commit: ae44f2177fa75c271734d5963972faecc3686c12 ("[RFC v2 2/2] sched/fair: introduce sched-idle balance") url: https://github.com/intel-lab-lkp/linux/commits/Abel-Wu/introduece-sched-idle-balance/20220409-215303 base: https://git.kernel.org/cgit/linux/kernel/git/tip/tip.git 089c02ae2771a14af2928c59c56abfb9b885a8d7 patch link: https://lore.kernel.org/lkml/20220409135104.3733193-3-wuyun.abel@bytedance.com in testcase: reaim on test machine: 16 threads 1 sockets Intel(R) Xeon(R) E-2278G CPU @ 3.40GHz with 32G memory with following parameters: runtime: 300s nr_task: 100% test: new_fserver cpufreq_governor: performance ucode: 0xec test-description: REAIM is an updated and improved version of AIM 7 benchmark. test-url: https://sourceforge.net/projects/re-aim-7/ Details are as below: --------------------------------------------------------------------------------------------------> To reproduce: git clone https://github.com/intel/lkp-tests.git cd lkp-tests sudo bin/lkp install job.yaml # job file is attached in this email bin/lkp split-job --compatible job.yaml # generate the yaml file for lkp run sudo bin/lkp run generated-yaml-file # if come across any failure that blocks the test, # please remove ~/.lkp and /lkp dir to run from a clean state. ========================================================================================= compiler/cpufreq_governor/kconfig/nr_task/rootfs/runtime/tbox_group/test/testcase/ucode: gcc-11/performance/x86_64-rhel-8.3/100%/debian-10.4-x86_64-20200603.cgz/300s/lkp-cfl-e1/new_fserver/reaim/0xec commit: 6b433275e3 ("sched/fair: filter out overloaded cpus in SIS") ae44f2177f ("sched/fair: introduce sched-idle balance") 6b433275e3a3cf18 ae44f2177fa75c271734d596397 ---------------- --------------------------- %stddev %change %stddev \ | \ 356517 +2.3% 364606 reaim.jobs_per_min 22282 +2.3% 22787 reaim.jobs_per_min_child 0.26 -2.2% 0.26 reaim.parent_time 1.87 ? 3% -33.1% 1.25 ? 4% reaim.std_dev_percent 0.00 ? 11% -83.0% 0.00 ? 22% reaim.std_dev_time 71394 +5.5% 75296 reaim.time.involuntary_context_switches 412752 +2.9% 424628 reaim.time.voluntary_context_switches 7647401 ? 24% -39.7% 4609658 cpuidle..usage 27840 ? 21% -34.3% 18280 vmstat.system.in 3736 +4.2% 3894 proc-vmstat.nr_active_anon 6959 +2.2% 7111 proc-vmstat.nr_shmem 3736 +4.2% 3894 proc-vmstat.nr_zone_active_anon 0.33 ? 18% -23.8% 0.25 ? 8% sched_debug.cfs_rq:/.h_nr_running.avg 0.31 ? 14% -20.7% 0.24 ? 7% sched_debug.cfs_rq:/.nr_running.avg 0.00 ? 25% -31.1% 0.00 ? 36% sched_debug.cpu.next_balance.stddev 3.20 ? 4% +0.5 3.71 ? 8% perf-stat.i.cache-miss-rate% 1818 ? 3% -10.2% 1632 ? 6% perf-stat.i.cycles-between-cache-misses 987044 ? 2% -3.8% 949746 perf-stat.i.iTLB-load-misses 442185 ? 8% -10.4% 395984 perf-stat.i.iTLB-loads 5920 ? 3% +3.8% 6144 perf-stat.i.instructions-per-iTLB-miss 0.05 ? 13% -0.0 0.05 perf-stat.overall.dTLB-load-miss-rate% 8510 ? 2% +3.6% 8814 perf-stat.overall.instructions-per-iTLB-miss 983682 ? 2% -3.8% 946402 perf-stat.ps.iTLB-load-misses 440691 ? 8% -10.5% 394589 perf-stat.ps.iTLB-loads 345730 ? 80% -92.5% 25933 ? 7% turbostat.C3 1.73 ? 86% -1.7 0.07 ? 7% turbostat.C3% 5033041 ? 31% -19.9% 4033568 turbostat.C6 59141 ?141% +395.4% 292968 ? 12% turbostat.C8 1.16 ?141% +4.9 6.03 ? 13% turbostat.C8% 24.98 ? 22% -29.1% 17.71 turbostat.CPU%c1 1.14 ? 84% -95.9% 0.05 ? 9% turbostat.CPU%c3 0.66 ?141% +540.9% 4.25 ? 15% turbostat.CPU%c7 20.56 ? 5% -8.4% 18.84 turbostat.CorWatt 8504754 ? 21% -34.5% 5573672 turbostat.IRQ 36.52 ? 29% +31.0% 47.86 turbostat.Pkg%pc3 21.00 ? 5% -8.3% 19.24 turbostat.PkgWatt 13.92 ? 7% -4.9 9.06 ? 2% perf-profile.calltrace.cycles-pp.secondary_startup_64_no_verify 13.08 ? 5% -4.7 8.36 ? 3% perf-profile.calltrace.cycles-pp.do_idle.cpu_startup_entry.secondary_startup_64_no_verify 13.09 ? 5% -4.7 8.37 ? 3% perf-profile.calltrace.cycles-pp.cpu_startup_entry.secondary_startup_64_no_verify 12.92 ? 5% -4.7 8.24 ? 3% perf-profile.calltrace.cycles-pp.cpuidle_idle_call.do_idle.cpu_startup_entry.secondary_startup_64_no_verify 12.45 ? 6% -4.4 8.01 ? 5% perf-profile.calltrace.cycles-pp.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle.cpu_startup_entry 11.98 ? 4% -4.3 7.72 ? 3% perf-profile.calltrace.cycles-pp.cpuidle_enter.cpuidle_idle_call.do_idle.cpu_startup_entry.secondary_startup_64_no_verify 9.04 ? 9% -3.5 5.53 ? 5% perf-profile.calltrace.cycles-pp.intel_idle.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle 9.00 ? 9% -3.5 5.52 ? 5% perf-profile.calltrace.cycles-pp.mwait_idle_with_hints.intel_idle.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call 2.68 ? 4% -0.8 1.88 ? 6% perf-profile.calltrace.cycles-pp.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle 2.40 ? 2% -0.7 1.74 ? 6% perf-profile.calltrace.cycles-pp.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call 2.34 ? 21% -0.6 1.72 ? 17% perf-profile.calltrace.cycles-pp.call_console_drivers.console_unlock.vprintk_emit.devkmsg_emit.devkmsg_write.cold 2.34 ? 21% -0.6 1.72 ? 17% perf-profile.calltrace.cycles-pp.console_unlock.vprintk_emit.devkmsg_emit.devkmsg_write.cold.new_sync_write 2.49 ? 11% -0.6 1.90 ? 11% perf-profile.calltrace.cycles-pp.drm_fb_helper_damage_blit_real.drm_fb_helper_damage_work.process_one_work.worker_thread.kthread 2.50 ? 11% -0.6 1.90 ? 11% perf-profile.calltrace.cycles-pp.drm_fb_helper_damage_work.process_one_work.worker_thread.kthread.ret_from_fork 2.52 ? 11% -0.6 1.93 ? 10% perf-profile.calltrace.cycles-pp.process_one_work.worker_thread.kthread.ret_from_fork 2.44 ? 11% -0.6 1.85 ? 11% perf-profile.calltrace.cycles-pp.memcpy_toio.drm_fb_helper_damage_blit_real.drm_fb_helper_damage_work.process_one_work.worker_thread 2.52 ? 11% -0.6 1.94 ? 10% perf-profile.calltrace.cycles-pp.worker_thread.kthread.ret_from_fork 2.97 ? 11% -0.6 2.40 ? 9% perf-profile.calltrace.cycles-pp.ret_from_fork 2.96 ? 11% -0.6 2.40 ? 9% perf-profile.calltrace.cycles-pp.kthread.ret_from_fork 1.61 ? 3% -0.5 1.11 ? 5% perf-profile.calltrace.cycles-pp.__sysvec_apic_timer_interrupt.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state.cpuidle_enter 1.49 ? 3% -0.4 1.05 ? 6% perf-profile.calltrace.cycles-pp.hrtimer_interrupt.__sysvec_apic_timer_interrupt.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state 1.14 ? 5% -0.3 0.87 ? 4% perf-profile.calltrace.cycles-pp.__hrtimer_run_queues.hrtimer_interrupt.__sysvec_apic_timer_interrupt.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt 4.69 ? 3% +0.3 5.00 perf-profile.calltrace.cycles-pp.div_long 0.90 ? 29% +0.4 1.34 ? 7% perf-profile.calltrace.cycles-pp.copy_process.kernel_clone.__do_sys_clone.do_syscall_64.entry_SYSCALL_64_after_hwframe 12.40 ? 5% +1.4 13.85 ? 2% perf-profile.calltrace.cycles-pp.string_rtns_1 13.92 ? 7% -4.9 9.06 ? 2% perf-profile.children.cycles-pp.do_idle 13.92 ? 7% -4.9 9.06 ? 2% perf-profile.children.cycles-pp.secondary_startup_64_no_verify 13.92 ? 7% -4.9 9.06 ? 2% perf-profile.children.cycles-pp.cpu_startup_entry 13.76 ? 7% -4.8 8.94 ? 2% perf-profile.children.cycles-pp.cpuidle_idle_call 12.74 ? 6% -4.4 8.37 ? 3% perf-profile.children.cycles-pp.cpuidle_enter_state 12.74 ? 6% -4.4 8.37 ? 3% perf-profile.children.cycles-pp.cpuidle_enter 9.14 ? 7% -3.4 5.73 ? 3% perf-profile.children.cycles-pp.intel_idle 9.10 ? 7% -3.4 5.71 ? 3% perf-profile.children.cycles-pp.mwait_idle_with_hints 3.42 ? 5% -0.8 2.58 ? 6% perf-profile.children.cycles-pp.asm_sysvec_apic_timer_interrupt 3.02 ? 3% -0.7 2.34 ? 6% perf-profile.children.cycles-pp.sysvec_apic_timer_interrupt 2.49 ? 11% -0.6 1.90 ? 11% perf-profile.children.cycles-pp.memcpy_toio 2.49 ? 11% -0.6 1.90 ? 11% perf-profile.children.cycles-pp.drm_fb_helper_damage_blit_real 2.50 ? 11% -0.6 1.90 ? 11% perf-profile.children.cycles-pp.drm_fb_helper_damage_work 2.52 ? 11% -0.6 1.93 ? 10% perf-profile.children.cycles-pp.process_one_work 2.52 ? 11% -0.6 1.94 ? 10% perf-profile.children.cycles-pp.worker_thread 2.96 ? 11% -0.6 2.40 ? 9% perf-profile.children.cycles-pp.kthread 3.00 ? 11% -0.6 2.44 ? 9% perf-profile.children.cycles-pp.ret_from_fork 1.99 ? 2% -0.5 1.45 ? 5% perf-profile.children.cycles-pp.__sysvec_apic_timer_interrupt 1.85 -0.5 1.38 ? 5% perf-profile.children.cycles-pp.hrtimer_interrupt 0.81 ? 19% -0.4 0.44 ? 7% perf-profile.children.cycles-pp.menu_select 1.44 ? 3% -0.3 1.14 ? 5% perf-profile.children.cycles-pp.__hrtimer_run_queues 0.41 ? 5% -0.1 0.32 ? 14% perf-profile.children.cycles-pp.io_serial_out 0.12 ? 27% -0.1 0.06 ? 36% perf-profile.children.cycles-pp.rcu_idle_exit 0.16 ? 16% -0.1 0.10 ? 17% perf-profile.children.cycles-pp.clockevents_program_event 0.09 ? 25% -0.1 0.04 ? 69% perf-profile.children.cycles-pp.native_apic_msr_eoi_write 0.14 ? 22% -0.0 0.09 ? 13% perf-profile.children.cycles-pp.enqueue_hrtimer 0.15 ? 10% -0.0 0.11 ? 19% perf-profile.children.cycles-pp.sched_clock_cpu 0.08 ? 14% -0.0 0.04 ? 69% perf-profile.children.cycles-pp.cpuidle_governor_latency_req 0.12 ? 23% -0.0 0.08 ? 15% perf-profile.children.cycles-pp.timerqueue_add 4.70 ? 3% +0.3 5.00 perf-profile.children.cycles-pp.div_long 12.46 ? 5% +1.5 13.91 ? 2% perf-profile.children.cycles-pp.string_rtns_1 9.10 ? 7% -3.4 5.71 ? 3% perf-profile.self.cycles-pp.mwait_idle_with_hints 2.46 ? 11% -0.6 1.84 ? 10% perf-profile.self.cycles-pp.memcpy_toio 0.42 ? 29% -0.2 0.20 ? 13% perf-profile.self.cycles-pp.menu_select 0.40 ? 23% -0.2 0.24 ? 15% perf-profile.self.cycles-pp.cpuidle_enter_state 0.41 ? 5% -0.1 0.32 ? 14% perf-profile.self.cycles-pp.io_serial_out 0.09 ? 25% -0.1 0.04 ? 69% perf-profile.self.cycles-pp.native_apic_msr_eoi_write 0.08 ? 19% -0.0 0.04 ? 52% perf-profile.self.cycles-pp.cpuidle_idle_call 4.67 ? 3% +0.3 4.98 perf-profile.self.cycles-pp.div_long 12.36 ? 5% +1.4 13.78 ? 2% perf-profile.self.cycles-pp.string_rtns_1 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://01.org/lkp