Greeting, There is no primary kpi change in this test, below is the data collected through multiple monitors running background just for your information. commit d4550809586dc011ef8c5a9179e6a4325e7b98e1 ("sched: look for idle cpu at wake up") https://git.linaro.org/people/vincent.guittot/kernel.git sched/pelt in testcase: hackbench on test machine: 8 threads Ivy Bridge with 16G memory with following parameters: nr_threads: 50% mode: process ipc: socket cpufreq_governor: performance test-description: Hackbench is both a benchmark and a stress test for the Linux kernel scheduler. test-url: https://github.com/linux-test-project/ltp/blob/master/testcases/kernel/sc... 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/socket/x86_64-rhel-7.2/process/50%/debian-x86_64-2016-08-31.cgz/lkp-ivb-d01/hackbench commit: fba9fb5f5f ("sched: use load_avg for selecting idlest group") d455080958 ("sched: look for idle cpu at wake up") fba9fb5f5f79866e d4550809586dc011ef8c5a9179 ---------------- -------------------------- %stddev %change %stddev \ | \ 34162896 ± 1% -8.0% 31437510 ± 1% hackbench.time.involuntary_context_switches 733.00 ± 0% +1.3% 742.25 ± 0% hackbench.time.percent_of_cpu_this_job_got 1.547e+08 ± 1% -9.7% 1.397e+08 ± 0% hackbench.time.voluntary_context_switches 20736 ± 20% +254.5% 73507 ± 25% softirqs.NET_RX 432890 ± 15% +89.1% 818743 ± 18% softirqs.RCU 528670 ± 0% +18.9% 628410 ± 1% softirqs.SCHED 75.25 ± 1% -11.3% 66.75 ± 0% vmstat.procs.r 363821 ± 0% -6.3% 341078 ± 0% vmstat.system.cs 56805 ± 0% -7.4% 52594 ± 0% vmstat.system.in 6352414 ± 2% -17.8% 5221427 ± 1% proc-vmstat.numa_hit 6352382 ± 2% -17.8% 5221393 ± 1% proc-vmstat.numa_local 18746436 ± 1% -18.5% 15281493 ± 1% proc-vmstat.pgalloc_normal 18734627 ± 1% -18.5% 15269275 ± 1% proc-vmstat.pgfree 92.92 ± 0% +1.3% 94.16 ± 0% turbostat.%Busy 3431 ± 0% +1.3% 3476 ± 0% turbostat.Avg_MHz 6.95 ± 0% -17.6% 5.72 ± 0% turbostat.CPU%c1 0.03 ± 0% -50.0% 0.02 ± 33% turbostat.CPU%c3 49520562 ± 3% -47.6% 25951758 ± 1% cpuidle.C1E-IVB.time 874280 ± 1% -17.1% 725099 ± 1% cpuidle.C1E-IVB.usage 24613560 ± 5% -78.2% 5355682 ± 1% cpuidle.C3-IVB.time 196895 ± 3% -63.3% 72279 ± 1% cpuidle.C3-IVB.usage 12706889 ± 5% -45.6% 6918345 ± 1% cpuidle.C6-IVB.time 37427 ± 8% -75.0% 9356 ± 1% cpuidle.C6-IVB.usage 181016 ± 1% +21.0% 218998 ± 2% cpuidle.POLL.usage 0.50 ± 0% +1.7% 0.51 ± 0% perf-stat.branch-miss-rate% 10.45 ± 3% -16.3% 8.75 ± 1% perf-stat.cache-miss-rate% 1.051e+11 ± 1% +13.1% 1.189e+11 ± 1% perf-stat.cache-references 2.227e+08 ± 1% -6.8% 2.075e+08 ± 0% perf-stat.context-switches 15827085 ± 0% +55.5% 24608395 ± 1% perf-stat.cpu-migrations 2.82e+12 ± 1% -2.5% 2.749e+12 ± 1% perf-stat.dTLB-stores 9.469e+08 ± 1% -3.6% 9.126e+08 ± 1% perf-stat.iTLB-load-misses 0.74 ± 0% -2.7% 0.72 ± 1% perf-stat.ipc 16738 ± 11% -31.2% 11515 ± 16% sched_debug.cfs_rq:/.min_vruntime.stddev 16747 ± 11% -31.3% 11512 ± 16% sched_debug.cfs_rq:/.spread0.stddev 1107 ± 4% +11.0% 1229 ± 3% sched_debug.cfs_rq:/.util_avg.avg 1497 ± 6% +12.4% 1683 ± 3% sched_debug.cfs_rq:/.util_avg.max 239.41 ± 11% +19.2% 285.40 ± 5% sched_debug.cfs_rq:/.util_avg.stddev 226803 ± 7% -34.6% 148369 ± 8% sched_debug.cpu.avg_idle.max 69458 ± 8% -41.4% 40697 ± 35% sched_debug.cpu.avg_idle.stddev 1518435 ± 2% +16.3% 1766592 ± 6% sched_debug.cpu.nr_switches.stddev 3.33 ± 2% -8.0% 3.06 ± 1% perf-profile.calltrace.cycles-pp.__kmalloc_reserve.__alloc_skb.alloc_skb_with_frags.sock_alloc_send_pskb.unix_stream_sendmsg 3.02 ± 5% +16.7% 3.52 ± 2% perf-profile.calltrace.cycles-pp.__wake_up_common.__wake_up_sync_key.sock_def_readable.unix_stream_sendmsg.sock_sendmsg 4.20 ± 6% +26.3% 5.30 ± 2% perf-profile.calltrace.cycles-pp.__wake_up_sync_key.sock_def_readable.unix_stream_sendmsg.sock_sendmsg.sock_write_iter 1.01 ± 12% +38.0% 1.39 ± 9% perf-profile.calltrace.cycles-pp._raw_spin_lock.sock_sendmsg.sock_write_iter.__vfs_write.vfs_write 0.77 ± 21% +114.2% 1.66 ± 2% perf-profile.calltrace.cycles-pp._raw_spin_lock_irqsave.__wake_up_sync_key.sock_def_readable.unix_stream_sendmsg.sock_sendmsg 2.92 ± 5% +17.5% 3.43 ± 2% perf-profile.calltrace.cycles-pp.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.sock_def_readable.unix_stream_sendmsg 4.46 ± 2% -18.0% 3.65 ± 1% perf-profile.calltrace.cycles-pp.call_cpuidle.cpu_startup_entry.start_secondary 1.05 ± 9% -22.4% 0.82 ± 7% perf-profile.calltrace.cycles-pp.copy_user_enhanced_fast_string.skb_copy_datagram_from_iter.unix_stream_sendmsg.sock_sendmsg.sock_write_iter 5.43 ± 2% -13.8% 4.68 ± 1% perf-profile.calltrace.cycles-pp.cpu_startup_entry.start_secondary 4.45 ± 2% -18.1% 3.65 ± 1% perf-profile.calltrace.cycles-pp.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary 4.42 ± 2% -18.4% 3.61 ± 1% perf-profile.calltrace.cycles-pp.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary 2.88 ± 5% +17.2% 3.38 ± 2% perf-profile.calltrace.cycles-pp.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.sock_def_readable 4.78 ± 3% -25.7% 3.55 ± 1% perf-profile.calltrace.cycles-pp.intel_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry 0.71 ± 22% +116.1% 1.54 ± 2% perf-profile.calltrace.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock_irqsave.__wake_up_sync_key.sock_def_readable.unix_stream_sendmsg 0.00 ± -1% +Inf% 0.82 ± 22% perf-profile.calltrace.cycles-pp.select_task_rq_fair.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common 1.77 ± 7% +15.5% 2.04 ± 2% perf-profile.calltrace.cycles-pp.skb_queue_tail.unix_stream_sendmsg.sock_sendmsg.sock_write_iter.__vfs_write 5.76 ± 4% +22.9% 7.07 ± 1% perf-profile.calltrace.cycles-pp.sock_def_readable.unix_stream_sendmsg.sock_sendmsg.sock_write_iter.__vfs_write 5.47 ± 2% -13.8% 4.71 ± 1% perf-profile.calltrace.cycles-pp.start_secondary 2.79 ± 5% +18.4% 3.31 ± 3% perf-profile.calltrace.cycles-pp.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key 1.63 ± 2% -13.6% 1.41 ± 1% perf-profile.children.cycles-pp.___slab_alloc 1.79 ± 2% -14.0% 1.54 ± 1% perf-profile.children.cycles-pp.__slab_alloc 3.47 ± 5% +18.1% 4.10 ± 1% perf-profile.children.cycles-pp.__wake_up_common 4.66 ± 5% +26.3% 5.89 ± 1% perf-profile.children.cycles-pp.__wake_up_sync_key 3.78 ± 4% +34.6% 5.08 ± 1% perf-profile.children.cycles-pp._raw_spin_lock_irqsave 3.37 ± 4% +18.9% 4.00 ± 2% perf-profile.children.cycles-pp.autoremove_wake_function 5.01 ± 1% -17.8% 4.12 ± 1% perf-profile.children.cycles-pp.call_cpuidle 6.11 ± 1% -13.4% 5.29 ± 0% perf-profile.children.cycles-pp.cpu_startup_entry 5.00 ± 1% -17.7% 4.11 ± 1% perf-profile.children.cycles-pp.cpuidle_enter 4.96 ± 1% -18.0% 4.07 ± 1% perf-profile.children.cycles-pp.cpuidle_enter_state 3.32 ± 5% +18.5% 3.93 ± 2% perf-profile.children.cycles-pp.default_wake_function 4.89 ± 1% -18.2% 4.00 ± 1% perf-profile.children.cycles-pp.intel_idle 2.36 ± 5% +47.3% 3.48 ± 4% perf-profile.children.cycles-pp.native_queued_spin_lock_slowpath 0.76 ± 6% +54.4% 1.18 ± 4% perf-profile.children.cycles-pp.prepare_to_wait 0.69 ± 6% +89.2% 1.31 ± 4% perf-profile.children.cycles-pp.select_task_rq_fair 5.80 ± 4% +22.7% 7.11 ± 1% perf-profile.children.cycles-pp.sock_def_readable 5.47 ± 2% -13.8% 4.71 ± 1% perf-profile.children.cycles-pp.start_secondary 3.21 ± 4% +19.8% 3.85 ± 2% perf-profile.children.cycles-pp.try_to_wake_up 1.48 ± 2% +10.7% 1.64 ± 3% perf-profile.children.cycles-pp.unix_write_space 2.44 ± 1% +16.1% 2.83 ± 1% perf-profile.self.cycles-pp._raw_spin_lock_irqsave 4.89 ± 1% -18.2% 4.00 ± 1% perf-profile.self.cycles-pp.intel_idle 2.36 ± 5% +47.3% 3.48 ± 4% perf-profile.self.cycles-pp.native_queued_spin_lock_slowpath 1.58 ± 0% +13.3% 1.79 ± 3% perf-profile.self.cycles-pp.sock_def_readable perf-stat.cpu-migrations 2.8e+07 ++----------------------------------------------------------------+ | O | 2.6e+07 O+ O O O O O O O O O O O O O O O O O O | | O O O O 2.4e+07 ++ O | | | 2.2e+07 ++ | | | 2e+07 ++ | | | 1.8e+07 ++ | | | 1.6e+07 *+.*..*.*..*..*..*.*..*..*..*.*..*..*..*.*..*..*..*.*..*..*..* | | | 1.4e+07 ++----------------------------------------------------------------+ turbostat.Avg_MHz 3490 ++-------------------------------------------------------------------+ | O O O O O O O | 3480 O+ O O O O O O O O O O | | O O O O O O O | | 3470 ++ | | | 3460 ++ | | | 3450 ++ | | | | | 3440 ++ .*..*.*..*..*.. .*..*..*..*..*.. | *..*..*. *..*..*..*..* *.. | 3430 ++-----------------------------------------------------*--*-*--*-----+ turbostat._Busy 94.6 ++-------------------------------------------------------------------+ | O | 94.4 O+ O O O O O O O O O O O O O O O O | 94.2 ++ O O O O O | | O O 94 ++ | 93.8 ++ | | | 93.6 ++ | 93.4 ++ | | | 93.2 ++ .*.. .*..*.. .*.. | 93 *+.*..*..*..*.*. *..*..*..*..*..* *. *..*.. | | *..*.*..* | 92.8 ++-------------------------------------------------------------------+ hackbench.time.percent_of_cpu_this_job_got 746 ++--------------------------------------------------------------------+ | O | 744 O+ O O O O O O O O O O O O O O O | | O O O O O | 742 ++ O O O | | 740 ++ | | | 738 ++ | | | 736 ++ | *.. .*..*.. .*..*.*..*.. .*..*..*..*..*.. | 734 ++ *. *. *..*..*..*. *. | | *..*..*..* | 732 ++--------------------------------------------------------------------+ [*] bisect-good sample [O] bisect-bad sample 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. Thanks, Xiaolong