FYI, we noticed a 9.9% improvement of hackbench.throughput due to commit: commit 3fa6c507319c897598512da91c010a4ad2ed682c ("mm: optimize copy_page_to/from_iter_iovec") https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master in testcase: hackbench on test machine: 12 threads Westmere with 6G memory with following parameters: nr_threads: 50% mode: process ipc: pipe cpufreq_governor: performance 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. Details are as below: --------------------------------------------------------------------------------------------------> To reproduce: git clone git://git.kernel.org/pub/scm/linux/kernel/git/wfg/lkp-tests.git cd lkp-tests bin/lkp install job.yaml # job file is attached in this email bin/lkp run job.yaml ========================================================================================= compiler/cpufreq_governor/ipc/kconfig/mode/nr_threads/rootfs/tbox_group/testcase: gcc-6/performance/pipe/x86_64-rhel/process/50%/debian-x86_64-2015-02-07.cgz/wsm/hackbench commit: 7e4411bfe6 ("mm: add cond_resched() to generic_swapfile_activate()") 3fa6c50731 ("mm: optimize copy_page_to/from_iter_iovec") 7e4411bfe6dd2971 3fa6c507319c897598512da91c ---------------- -------------------------- fail:runs %reproduction fail:runs | | | %stddev %change %stddev \ | \ 54186 ± 0% +9.9% 59563 ± 0% hackbench.throughput 623.16 ± 0% -1.4% 614.17 ± 0% hackbench.time.elapsed_time 623.16 ± 0% -1.4% 614.17 ± 0% hackbench.time.elapsed_time.max 2.558e+08 ± 1% -49.9% 1.28e+08 ± 0% hackbench.time.involuntary_context_switches 219479 ± 0% +8.3% 237691 ± 0% hackbench.time.minor_page_faults 1182 ± 0% -3.3% 1142 ± 0% hackbench.time.percent_of_cpu_this_job_got 6906 ± 0% -4.3% 6610 ± 0% hackbench.time.system_time 462.37 ± 0% -11.5% 409.01 ± 0% hackbench.time.user_time 9.813e+08 ± 1% -19.0% 7.953e+08 ± 0% hackbench.time.voluntary_context_switches 78641 ± 1% -1.5% 77430 ± 0% interrupts.CAL:Function_call_interrupts 283030 ± 4% +204.8% 862697 ± 0% softirqs.SCHED 99.14 ± 0% -1.7% 97.48 ± 0% turbostat.%Busy 3483 ± 0% -1.7% 3424 ± 0% turbostat.Avg_MHz 0.71 ± 3% +234.5% 2.38 ± 1% turbostat.CPU%c1 142.75 ± 1% -22.1% 111.25 ± 0% vmstat.procs.r 2020719 ± 1% -17.8% 1661290 ± 0% vmstat.system.cs 397397 ± 0% -41.9% 230705 ± 0% vmstat.system.in 35891846 ± 0% +28.6% 46143986 ± 0% proc-vmstat.numa_hit 35891843 ± 0% +28.6% 46143984 ± 0% proc-vmstat.numa_local 35957085 ± 0% +28.5% 46213934 ± 0% proc-vmstat.pgalloc_normal 35940003 ± 0% +28.5% 46198964 ± 0% proc-vmstat.pgfree 50485763 ± 3% +250.5% 1.769e+08 ± 1% cpuidle.C1-NHM.time 14312245 ± 3% +314.8% 59367078 ± 0% cpuidle.C1-NHM.usage 590743 ± 2% +118.0% 1287652 ± 5% cpuidle.C1E-NHM.time 13679 ± 4% +181.8% 38554 ± 4% cpuidle.C1E-NHM.usage 7545462 ± 5% +9.8% 8287020 ± 5% cpuidle.C3-NHM.time 20219 ± 3% +45.2% 29359 ± 4% cpuidle.C3-NHM.usage 2157694 ± 19% +242.3% 7386617 ± 5% cpuidle.POLL.time 329048 ± 3% +319.8% 1381417 ± 0% cpuidle.POLL.usage 2.656e+12 ± 0% +4.9% 2.785e+12 ± 0% perf-stat.branch-instructions 1.52 ± 0% -11.2% 1.35 ± 0% perf-stat.branch-miss-rate 4.044e+10 ± 0% -6.9% 3.766e+10 ± 0% perf-stat.branch-misses 0.20 ± 1% +46.2% 0.29 ± 1% perf-stat.cache-miss-rate 3.172e+08 ± 1% +23.4% 3.913e+08 ± 1% perf-stat.cache-misses 1.622e+11 ± 1% -15.7% 1.368e+11 ± 0% perf-stat.cache-references 1.264e+09 ± 1% -18.9% 1.025e+09 ± 0% perf-stat.context-switches 2.588e+13 ± 0% -3.1% 2.508e+13 ± 0% perf-stat.cpu-cycles 7197359 ± 1% +316.2% 29954541 ± 2% perf-stat.cpu-migrations 4.154e+12 ± 0% +4.4% 4.338e+12 ± 0% perf-stat.dTLB-loads 2.551e+12 ± 0% +4.7% 2.67e+12 ± 0% perf-stat.dTLB-stores 0.03 ± 1% -26.7% 0.02 ± 0% perf-stat.iTLB-load-miss-rate 4.549e+09 ± 1% -23.6% 3.477e+09 ± 0% perf-stat.iTLB-load-misses 1.401e+13 ± 0% +4.2% 1.46e+13 ± 0% perf-stat.iTLB-loads 1.401e+13 ± 0% +4.3% 1.46e+13 ± 0% perf-stat.instructions 3079 ± 1% +36.3% 4199 ± 0% perf-stat.instructions-per-iTLB-miss 0.54 ± 0% +7.6% 0.58 ± 0% perf-stat.ipc 979420 ± 0% +8.8% 1065802 ± 0% perf-stat.minor-faults 1.094e+08 ± 1% +27.5% 1.395e+08 ± 1% perf-stat.node-loads 1.952e+08 ± 1% +22.1% 2.383e+08 ± 0% perf-stat.node-stores 979409 ± 0% +8.8% 1065778 ± 0% perf-stat.page-faults 18834 ± 22% -65.7% 6451 ± 31% sched_debug.cfs_rq:/.load.min 5653832 ± 1% +25.9% 7116466 ± 0% sched_debug.cfs_rq:/.min_vruntime.avg 5731022 ± 1% +26.1% 7228446 ± 0% sched_debug.cfs_rq:/.min_vruntime.max 5588870 ± 1% +25.3% 7000867 ± 0% sched_debug.cfs_rq:/.min_vruntime.min 41841 ± 6% +74.3% 72909 ± 8% sched_debug.cfs_rq:/.min_vruntime.stddev 0.14 ± 24% +53.0% 0.21 ± 9% sched_debug.cfs_rq:/.nr_running.stddev 99.27 ± 2% -13.9% 85.52 ± 2% sched_debug.cfs_rq:/.runnable_load_avg.max 39.55 ± 9% -25.9% 29.30 ± 14% sched_debug.cfs_rq:/.runnable_load_avg.min 4695 ±676% +1191.9% 60659 ± 67% sched_debug.cfs_rq:/.spread0.avg 81802 ± 33% +111.0% 172600 ± 26% sched_debug.cfs_rq:/.spread0.max 41801 ± 6% +74.3% 72845 ± 8% sched_debug.cfs_rq:/.spread0.stddev 7.50 ± 27% -44.4% 4.17 ± 13% sched_debug.cpu.clock.stddev 7.50 ± 27% -44.4% 4.17 ± 13% sched_debug.cpu.clock_task.stddev 68.60 ± 3% -19.0% 55.56 ± 5% sched_debug.cpu.cpu_load[0].avg 98.59 ± 1% -11.6% 87.11 ± 3% sched_debug.cpu.cpu_load[0].max 21.36 ± 16% +31.0% 27.99 ± 11% sched_debug.cpu.cpu_load[0].stddev 100.68 ± 3% -13.7% 86.89 ± 3% sched_debug.cpu.cpu_load[1].max 100.61 ± 3% -14.7% 85.86 ± 2% sched_debug.cpu.cpu_load[2].max 12.73 ± 10% -24.5% 9.62 ± 15% sched_debug.cpu.cpu_load[2].stddev 99.91 ± 3% -14.9% 85.02 ± 2% sched_debug.cpu.cpu_load[3].max 12.44 ± 11% -29.0% 8.83 ± 12% sched_debug.cpu.cpu_load[3].stddev 98.70 ± 3% -14.3% 84.64 ± 2% sched_debug.cpu.cpu_load[4].max 12.07 ± 11% -29.6% 8.50 ± 11% sched_debug.cpu.cpu_load[4].stddev 1514 ± 27% +59.5% 2415 ± 21% sched_debug.cpu.curr->pid.stddev 17382 ± 42% -76.5% 4082 ± 24% sched_debug.cpu.load.min 9.72 ± 10% -27.8% 7.02 ± 6% sched_debug.cpu.nr_running.avg 20.45 ± 8% -15.8% 17.23 ± 3% sched_debug.cpu.nr_running.max 2.66 ± 47% -80.3% 0.52 ± 22% sched_debug.cpu.nr_running.min 50594782 ± 1% -18.0% 41509427 ± 0% sched_debug.cpu.nr_switches.avg 53214998 ± 1% -16.7% 44306289 ± 0% sched_debug.cpu.nr_switches.max 48171349 ± 1% -19.0% 39022763 ± 0% sched_debug.cpu.nr_switches.min 0.13 ± 43% +1745.7% 2.45 ± 10% sched_debug.cpu.nr_uninterruptible.avg 265.23 ± 23% +584.3% 1814 ± 18% sched_debug.cpu.nr_uninterruptible.max -260.93 ±-26% +683.9% -2045 ± -7% sched_debug.cpu.nr_uninterruptible.min 147.42 ± 9% +719.8% 1208 ± 12% sched_debug.cpu.nr_uninterruptible.stddev 0.32 ±100% +320.9% 1.36 ± 5% perf-profile.cycles-pp.__fget_light.sys_write.entry_SYSCALL_64_fastpath 1.57 ± 16% +107.0% 3.25 ± 2% perf-profile.cycles-pp.__mutex_lock_slowpath.mutex_lock.pipe_write.__vfs_write.vfs_write 22.20 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.__read_nocancel 2.32 ± 7% -63.5% 0.85 ± 3% perf-profile.cycles-pp.__schedule.schedule.exit_to_usermode_loop.syscall_return_slowpath.entry_SYSCALL_64_fastpath 13.13 ± 1% -22.2% 10.21 ± 1% perf-profile.cycles-pp.__schedule.schedule.pipe_wait.pipe_read.__vfs_read 1.03 ± 17% +47.6% 1.53 ± 2% perf-profile.cycles-pp.__switch_to 12.11 ± 20% +104.5% 24.77 ± 0% perf-profile.cycles-pp.__vfs_read.vfs_read.sys_read.entry_SYSCALL_64_fastpath 14.31 ± 17% -100.0% 0.00 ± -1% perf-profile.cycles-pp.__vfs_read.vfs_read.sys_read.entry_SYSCALL_64_fastpath.__read_nocancel 19.46 ± 22% +132.0% 45.16 ± 0% perf-profile.cycles-pp.__vfs_write.vfs_write.sys_write.entry_SYSCALL_64_fastpath 25.25 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.__vfs_write.vfs_write.sys_write.entry_SYSCALL_64_fastpath.__write_nocancel 21.44 ± 2% -26.0% 15.87 ± 1% perf-profile.cycles-pp.__wake_up_common.__wake_up_sync_key.pipe_write.__vfs_write.vfs_write 29.77 ± 2% -11.1% 26.46 ± 1% perf-profile.cycles-pp.__wake_up_sync_key.pipe_write.__vfs_write.vfs_write.sys_write 33.50 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.__write_nocancel 1.62 ± 2% -29.2% 1.15 ± 6% perf-profile.cycles-pp._raw_spin_lock.__schedule.schedule.pipe_wait.pipe_read 4.84 ± 3% -27.2% 3.53 ± 3% perf-profile.cycles-pp._raw_spin_lock.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common 6.98 ± 5% +36.5% 9.53 ± 1% perf-profile.cycles-pp._raw_spin_lock_irqsave.__wake_up_sync_key.pipe_write.__vfs_write.vfs_write 7.36 ± 2% -26.0% 5.44 ± 1% perf-profile.cycles-pp.activate_task.ttwu_do_activate.try_to_wake_up.default_wake_function.autoremove_wake_function 20.76 ± 2% -26.9% 15.17 ± 1% perf-profile.cycles-pp.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.pipe_write.__vfs_write 0.00 ± -1% +Inf% 1.38 ± 2% perf-profile.cycles-pp.avc_has_perm.file_has_perm.selinux_file_permission.security_file_permission.rw_verify_area 0.94 ± 17% +216.7% 2.99 ± 3% perf-profile.cycles-pp.call_cpuidle.cpu_startup_entry.start_secondary 1.48 ± 4% -33.4% 0.99 ± 2% perf-profile.cycles-pp.check_preempt_curr.ttwu_do_wakeup.ttwu_do_activate.try_to_wake_up.default_wake_function 2.84 ± 5% +45.4% 4.14 ± 0% perf-profile.cycles-pp.copy_page_from_iter.pipe_write.__vfs_write.vfs_write.sys_write 0.86 ± 23% +129.1% 1.97 ± 1% perf-profile.cycles-pp.copy_page_from_iter_iovec.copy_page_from_iter.pipe_write.__vfs_write.vfs_write 3.85 ± 4% +28.8% 4.96 ± 1% perf-profile.cycles-pp.copy_page_to_iter.pipe_read.__vfs_read.vfs_read.sys_read 1.32 ± 22% +47.7% 1.96 ± 2% perf-profile.cycles-pp.copy_page_to_iter_iovec.copy_page_to_iter.pipe_read.__vfs_read.vfs_read 1.38 ± 13% +27.4% 1.76 ± 0% perf-profile.cycles-pp.copy_user_generic_string.copy_page_from_iter.pipe_write.__vfs_write.vfs_write 2.17 ± 5% +18.7% 2.57 ± 1% perf-profile.cycles-pp.copy_user_generic_string.copy_page_to_iter.pipe_read.__vfs_read.vfs_read 1.58 ± 16% +257.4% 5.65 ± 4% perf-profile.cycles-pp.cpu_startup_entry.start_secondary 0.93 ± 17% +217.1% 2.96 ± 3% perf-profile.cycles-pp.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary 0.93 ± 17% +215.6% 2.94 ± 3% perf-profile.cycles-pp.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary 5.60 ± 1% -18.2% 4.58 ± 1% perf-profile.cycles-pp.deactivate_task.__schedule.schedule.pipe_wait.pipe_read 20.53 ± 2% -27.0% 14.97 ± 2% perf-profile.cycles-pp.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.pipe_write 3.44 ± 1% -15.6% 2.90 ± 2% perf-profile.cycles-pp.dequeue_entity.dequeue_task_fair.deactivate_task.__schedule.schedule 4.66 ± 1% -17.3% 3.85 ± 1% perf-profile.cycles-pp.dequeue_task_fair.deactivate_task.__schedule.schedule.pipe_wait 5.33 ± 2% -26.9% 3.90 ± 2% perf-profile.cycles-pp.enqueue_entity.enqueue_task_fair.activate_task.ttwu_do_activate.try_to_wake_up 6.43 ± 2% -26.8% 4.71 ± 2% perf-profile.cycles-pp.enqueue_task_fair.activate_task.ttwu_do_activate.try_to_wake_up.default_wake_function 39.59 ± 22% +125.5% 89.25 ± 0% perf-profile.cycles-pp.entry_SYSCALL_64_fastpath 19.64 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.entry_SYSCALL_64_fastpath.__read_nocancel 31.99 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.entry_SYSCALL_64_fastpath.__write_nocancel 1.17 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.exit_to_usermode_loop.syscall_return_slowpath.entry_SYSCALL_64_fastpath.__write_nocancel 0.60 ± 16% +75.7% 1.05 ± 1% perf-profile.cycles-pp.file_has_perm.selinux_file_permission.security_file_permission.rw_verify_area.vfs_write 1.77 ± 6% -27.4% 1.29 ± 2% perf-profile.cycles-pp.idle_cpu.select_idle_sibling.select_task_rq_fair.try_to_wake_up.default_wake_function 0.89 ± 17% +215.8% 2.79 ± 3% perf-profile.cycles-pp.intel_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry 0.62 ± 15% +93.1% 1.20 ± 4% perf-profile.cycles-pp.mutex_lock.pipe_read.__vfs_read.vfs_read.sys_read 3.73 ± 1% +43.7% 5.36 ± 2% perf-profile.cycles-pp.mutex_lock.pipe_write.__vfs_write.vfs_write.sys_write 1.51 ± 16% +103.2% 3.06 ± 2% perf-profile.cycles-pp.mutex_optimistic_spin.__mutex_lock_slowpath.mutex_lock.pipe_write.__vfs_write 0.00 ± -1% +Inf% 1.01 ± 3% perf-profile.cycles-pp.mutex_spin_on_owner.isra.1.mutex_optimistic_spin.__mutex_lock_slowpath.mutex_lock.pipe_write 1.11 ± 20% +28.8% 1.43 ± 2% perf-profile.cycles-pp.mutex_unlock.__vfs_write.vfs_write.sys_write.entry_SYSCALL_64_fastpath 1.61 ± 2% -29.4% 1.14 ± 5% perf-profile.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock.__schedule.schedule.pipe_wait 4.82 ± 3% -27.1% 3.51 ± 3% perf-profile.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock.try_to_wake_up.default_wake_function.autoremove_wake_function 6.14 ± 6% +38.2% 8.49 ± 1% perf-profile.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock_irqsave.__wake_up_sync_key.pipe_write.__vfs_write 0.90 ± 26% -71.6% 0.26 ±100% perf-profile.cycles-pp.pick_next_task_fair.__schedule.schedule.exit_to_usermode_loop.syscall_return_slowpath 2.27 ± 2% -29.5% 1.60 ± 3% perf-profile.cycles-pp.pick_next_task_fair.__schedule.schedule.pipe_wait.pipe_read 15.47 ± 1% -21.2% 12.20 ± 1% perf-profile.cycles-pp.pipe_wait.pipe_read.__vfs_read.vfs_read.sys_read 1.60 ± 24% +197.5% 4.77 ± 1% perf-profile.cycles-pp.rw_verify_area.vfs_read.sys_read.entry_SYSCALL_64_fastpath 2.72 ± 15% -100.0% 0.00 ± -1% perf-profile.cycles-pp.rw_verify_area.vfs_read.sys_read.entry_SYSCALL_64_fastpath.__read_nocancel 1.46 ± 25% +181.8% 4.12 ± 0% perf-profile.cycles-pp.rw_verify_area.vfs_write.sys_write.entry_SYSCALL_64_fastpath 2.39 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.rw_verify_area.vfs_write.sys_write.entry_SYSCALL_64_fastpath.__write_nocancel 1.12 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.schedule.exit_to_usermode_loop.syscall_return_slowpath.entry_SYSCALL_64_fastpath.__write_nocancel 13.59 ± 1% -22.3% 10.57 ± 1% perf-profile.cycles-pp.schedule.pipe_wait.pipe_read.__vfs_read.vfs_read 3.57 ± 3% -26.9% 2.61 ± 1% perf-profile.cycles-pp.select_idle_sibling.select_task_rq_fair.try_to_wake_up.default_wake_function.autoremove_wake_function 4.55 ± 3% -28.6% 3.25 ± 1% perf-profile.cycles-pp.select_task_rq_fair.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common 1.60 ± 16% +257.4% 5.71 ± 4% perf-profile.cycles-pp.start_secondary 14.86 ± 21% +119.1% 32.55 ± 0% perf-profile.cycles-pp.sys_read.entry_SYSCALL_64_fastpath 18.81 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.sys_read.entry_SYSCALL_64_fastpath.__read_nocancel 22.39 ± 23% +138.6% 53.41 ± 0% perf-profile.cycles-pp.sys_write.entry_SYSCALL_64_fastpath 30.11 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.sys_write.entry_SYSCALL_64_fastpath.__write_nocancel 1.21 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.syscall_return_slowpath.entry_SYSCALL_64_fastpath.__write_nocancel 19.98 ± 2% -27.2% 14.55 ± 1% perf-profile.cycles-pp.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key 9.13 ± 2% -26.5% 6.71 ± 1% perf-profile.cycles-pp.ttwu_do_activate.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common 1.66 ± 3% -27.9% 1.20 ± 2% perf-profile.cycles-pp.ttwu_do_wakeup.ttwu_do_activate.try_to_wake_up.default_wake_function.autoremove_wake_function 1.37 ± 2% -15.4% 1.16 ± 1% perf-profile.cycles-pp.update_cfs_shares.dequeue_entity.dequeue_task_fair.deactivate_task.__schedule 1.83 ± 3% -26.4% 1.35 ± 2% perf-profile.cycles-pp.update_cfs_shares.enqueue_entity.enqueue_task_fair.activate_task.ttwu_do_activate 14.32 ± 20% +118.2% 31.25 ± 0% perf-profile.cycles-pp.vfs_read.sys_read.entry_SYSCALL_64_fastpath 18.00 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.vfs_read.sys_read.entry_SYSCALL_64_fastpath.__read_nocancel 21.62 ± 23% +137.5% 51.34 ± 0% perf-profile.cycles-pp.vfs_write.sys_write.entry_SYSCALL_64_fastpath 28.84 ± 16% -100.0% 0.00 ± -1% perf-profile.cycles-pp.vfs_write.sys_write.entry_SYSCALL_64_fastpath.__write_nocancel 0.91 ± 5% +113.2% 1.94 ± 2% perf-profile.func.cycles-pp.___might_sleep 1.00 ± 2% -47.2% 0.53 ± 4% perf-profile.func.cycles-pp.__enqueue_entity 1.05 ± 6% +101.9% 2.12 ± 3% perf-profile.func.cycles-pp.__might_sleep 2.29 ± 2% -23.9% 1.74 ± 3% perf-profile.func.cycles-pp.__schedule 2.53 ± 1% -23.8% 1.93 ± 2% perf-profile.func.cycles-pp.__switch_to 1.06 ± 1% -10.6% 0.95 ± 4% perf-profile.func.cycles-pp._raw_spin_lock 1.04 ± 2% -18.2% 0.85 ± 2% perf-profile.func.cycles-pp._raw_spin_unlock_irqrestore 1.19 ± 5% -13.1% 1.03 ± 1% perf-profile.func.cycles-pp.copy_page_from_iter_iovec 1.61 ± 4% -32.4% 1.09 ± 4% perf-profile.func.cycles-pp.copy_page_to_iter_iovec 3.77 ± 4% +15.3% 4.35 ± 0% perf-profile.func.cycles-pp.copy_user_generic_string 1.92 ± 6% -25.2% 1.44 ± 1% perf-profile.func.cycles-pp.idle_cpu 0.95 ± 17% +214.8% 2.98 ± 3% perf-profile.func.cycles-pp.intel_idle 0.65 ± 5% +51.0% 0.98 ± 4% perf-profile.func.cycles-pp.mutex_optimistic_spin 0.52 ± 3% +114.4% 1.11 ± 2% perf-profile.func.cycles-pp.mutex_spin_on_owner.isra.1 2.32 ± 4% +15.0% 2.67 ± 4% perf-profile.func.cycles-pp.pipe_write 1.04 ± 4% -24.1% 0.79 ± 3% perf-profile.func.cycles-pp.select_idle_sibling 1.44 ± 4% -28.3% 1.03 ± 4% perf-profile.func.cycles-pp.switch_mm_irqs_off 0.98 ± 3% -19.4% 0.79 ± 1% perf-profile.func.cycles-pp.update_cfs_rq_load_avg 1.58 ± 3% -17.9% 1.30 ± 1% perf-profile.func.cycles-pp.update_cfs_shares 2.45 ± 2% -26.9% 1.79 ± 2% perf-profile.func.cycles-pp.update_curr 1.07 ± 6% +18.1% 1.27 ± 3% perf-profile.func.cycles-pp.vfs_write perf-stat.cache-references 1.7e+11 ++---------------------------------------------------------------+ | | 1.65e+11 ++ * 1.6e+11 +* * * **.* .* *. * .* *. * * * .*| * + :*.***.*: : : * :: ***.** * :: ***.** + *.* * * .* * | 1.55e+11 ++ * : : :: * * ** + :* | | *.* : * | 1.5e+11 ++ * | | | 1.45e+11 ++ | 1.4e+11 ++ | | O O | 1.35e+11 OO O O O OOO OOO O O O OO | | OO OO OOO | 1.3e+11 ++---------------------------------------------------------------+ perf-stat.context-switches 1.3e+09 ++---------------------------------------------------------------* |* * * .* *. * .* *. * * | 1.25e+09 *++ :*.***.*: *.* * :: ***.** * :: ***.** : ** :*.*| | * : * : : * * : *.* : * .* | 1.2e+09 ++ *. : :: ** : :* | | * : * | 1.15e+09 ++ * | | | 1.1e+09 ++ | | | 1.05e+09 ++ | O O OOO OO | 1e+09 +O O O O OOO OOO OOO OOO | | OO | 9.5e+08 ++---------------------------------------------------------------+ perf-stat.cpu-migrations 3.5e+07 ++----------------------------------------------------------------+ | | 3e+07 ++ O O OO O O O O | OO OOO OO OOO O O O O O O O | | | 2.5e+07 ++ | | | 2e+07 ++ | | | 1.5e+07 ++ | | | | | 1e+07 ++ | **.* *.**.***.***. **.* *.**.***.***.***.** 5e+06 ++--**-***-**-------------***-***-*----**-***-**------------------+ turbostat.Avg_MHz 3490 ++-------------------------------------------------------------------+ | *.***. *.* .**.** .**. .**.** .**. .* .* .** 3480 **.* * **.**.**.*** * *** * **.***.** ** * | | | 3470 ++ | | | 3460 ++ | | | 3450 ++ | | | 3440 ++ | | | 3430 ++ O O OO | OO O OO OO OOO OO OOO OO OOO O | 3420 ++-------------------------------------------------------------------+ turbostat._Busy 99.2 ++--*-***--*-*-------------**-**--**-----**-**--**-------------------* **.* * **.**.* .*** * *** * **.***.**.***.**.*| 99 ++ * | 98.8 ++ | | | 98.6 ++ | 98.4 ++ | | | 98.2 ++ | 98 ++ | | | 97.8 ++ | 97.6 ++ | OO OO O O OO OOO OO OO OOO O O O | 97.4 ++-----O--------------------O--O--O----------------------------------+ turbostat.CPU_c1 2.6 ++--------------------------------------------------------------------+ 2.4 ++ O O O O | OO OO O O O OO OO OOO OO OO OO O | 2.2 ++ | 2 ++ | | | 1.8 ++ | 1.6 ++ | 1.4 ++ | | | 1.2 ++ | 1 ++ | | .* *. *. | 0.8 **.**.**.***.** *.** **.**.***.**.**.***.**.**.***.* **.***.**.**.*| 0.6 ++--------------------------------------------------------------------* perf-profile.cycles-pp.start_secondary 7 ++----------------------------------------------------------------------+ | | 6 +O OO | O O OO O O OO O O OO OO | 5 ++ O O O O OOO | | O | 4 ++ | | * | 3 ++ :: | |*. * * :: | 2 ++ * .* *.* .**.* :*. * .* : *. * *.**.: :*.* *.*| | * :.* .* * *.: * + *. * :: * + * .* * * *.**.* | 1 *+ * * * * * * * * * | | 0 ++----------------------------------------------------------------------+ [*] bisect-good sample [O] bisect-bad sample Thanks, Xiaolong