FYI, we noticed that turbostat.PkgWatt +13.7% change with your commit. https://github.com/0day-ci/linux Rik-van-Riel/cpuidle-break-out-of-idle-polling-loop-after-HLT-threshold/20160319-053854 commit 664ef5b8615ff12b03b46988408b4c0a15eab07c ("cpuidle: use high confidence factors only when considering polling") ========================================================================================= cluster/compiler/cpufreq_governor/ip/kconfig/nr_threads/rootfs/runtime/tbox_group/test/testcase: cs-localhost/gcc-4.9/performance/ipv4/x86_64-rhel/1/debian-x86_64-2015-02-07.cgz/300s/lkp-hsw-d01/SCTP_RR/netperf commit: 38497f04cd3604396327ba1810dde0ac239989f4 664ef5b8615ff12b03b46988408b4c0a15eab07c 38497f04cd360439 664ef5b8615ff12b03b4698840 ---------------- -------------------------- %stddev %change %stddev \ | \ 67771 ± 1% +2.8% 69694 ± 1% netperf.Throughput_tps 77.00 ± 0% +1.3% 78.00 ± 0% netperf.time.percent_of_cpu_this_job_got 226.75 ± 0% +3.3% 234.16 ± 0% netperf.time.system_time 20330477 ± 1% +2.8% 20907309 ± 1% netperf.time.voluntary_context_switches 9.29 ± 10% -42.2% 5.38 ± 6% time.user_time 266114 ± 1% +2.8% 273497 ± 1% vmstat.system.cs 1929 ± 1% +374.1% 9148 ± 23% vmstat.system.in 1.622e+08 ± 6% -64.3% 57961470 ± 6% cpuidle.C1-HSW.time 33127115 ± 3% -96.7% 1109711 ± 50% cpuidle.C1-HSW.usage 5649502 ± 10% +16.0% 6551668 ± 2% cpuidle.C3-HSW.time 63535957 ± 10% +117.7% 1.383e+08 ± 1% cpuidle.POLL.time 7659494 ± 10% +1097.5% 91718696 ± 3% cpuidle.POLL.usage 21.06 ± 1% +16.3% 24.50 ± 0% turbostat.%Busy 794.00 ± 2% +19.7% 950.33 ± 0% turbostat.Avg_MHz 20.14 ± 12% -20.2% 16.06 ± 10% turbostat.CPU%c1 0.38 ± 5% +18.5% 0.45 ± 3% turbostat.CPU%c3 17.24 ± 0% +21.0% 20.85 ± 2% turbostat.CorWatt 25.77 ± 0% +13.7% 29.31 ± 1% turbostat.PkgWatt 0.51 ± 39% +164.7% 1.36 ± 36% perf-profile.cycles-pp._raw_spin_lock_bh.lock_sock_nested.sctp_skb_recv_datagram.sctp_recvmsg.sock_common_recvmsg 0.51 ± 34% +142.9% 1.23 ± 27% perf-profile.cycles-pp.cpuidle_select.cpu_startup_entry.start_secondary 1.34 ± 7% +11.8% 1.50 ± 5% perf-profile.cycles-pp.dequeue_entity.dequeue_task_fair.deactivate_task.__schedule.schedule 14.74 ± 14% -78.4% 3.18 ± 68% perf-profile.cycles-pp.intel_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry 1.32 ± 12% -28.5% 0.94 ± 18% perf-profile.cycles-pp.is_ftrace_trampoline.print_context_stack.dump_trace.save_stack_trace_tsk.__account_scheduler_latency 1.16 ± 17% -21.3% 0.91 ± 19% perf-profile.cycles-pp.kfree_skb.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 0.00 ± -1% +Inf% 17.25 ± 4% perf-profile.cycles-pp.ktime_get.poll_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle 0.68 ± 31% +126.6% 1.55 ± 33% perf-profile.cycles-pp.lock_sock_nested.sctp_skb_recv_datagram.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg 0.38 ± 41% +178.9% 1.06 ± 28% perf-profile.cycles-pp.menu_select.cpuidle_select.cpu_startup_entry.start_secondary 0.30 ± 82% +304.4% 1.22 ± 42% perf-profile.cycles-pp.native_queued_spin_lock_slowpath._raw_spin_lock_bh.lock_sock_nested.sctp_skb_recv_datagram.sctp_recvmsg 9.13 ± 35% +165.1% 24.19 ± 3% perf-profile.cycles-pp.poll_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry 0.00 ± -1% +Inf% 10.46 ± 2% perf-profile.cycles-pp.read_tsc.ktime_get.poll_idle.cpuidle_enter_state.cpuidle_enter 1.14 ± 14% -16.2% 0.95 ± 8% perf-profile.cycles-pp.sctp_make_control.sctp_make_sack.sctp_packet_append_chunk.sctp_packet_transmit_chunk.sctp_outq_flush 1.71 ± 10% -16.3% 1.43 ± 5% perf-profile.cycles-pp.sctp_make_sack.sctp_packet_append_chunk.sctp_packet_transmit_chunk.sctp_outq_flush.sctp_outq_uncork 3.00 ± 8% -12.8% 2.62 ± 4% perf-profile.cycles-pp.sctp_packet_transmit_chunk.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.sctp_do_sm 1.40 ± 7% -23.9% 1.06 ± 11% perf-profile.cycles-pp.sctp_transport_reset_timers.sctp_outq_flush.sctp_outq_uncork.sctp_cmd_interpreter.sctp_do_sm 1.59 ± 20% -24.9% 1.19 ± 24% perf-profile.cycles-pp.sctp_ulpevent_free.sctp_recvmsg.sock_common_recvmsg.sock_recvmsg.___sys_recvmsg lkp-hsw-d01: Haswell Memory: 8G turbostat.PkgWatt 35 ++---------------------------------------------------------------------+ | O | 30 OOO OO O OOO O O | | O * * | 25 ** ******.*****.******.* ***.*** *.******.** **.******.*****.* ** *.*** |: : : : : : : : | 20 ++ : : : : : : : | |: : : : : : : : | 15 ++ : : : : : : : | | :: :: :: :: | 10 ++:: : : : | | :: : : : | 5 ++: : : : | | : : : : | 0 ++*------O----O----------*-------*-----------*-------------------------+ [*] bisect-good sample [O] bisect-bad sample 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 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 Ye.