[lkp] [x86 tsc_msr] 8c82d5a94b: unixbench.score -15.6% regression

From: kernel test robot <hidden>
Date: 2016-06-21 02:18:31
Also in: linux-acpi, lkml, oe-lkp

FYI, we noticed a -15.6% regression of unixbench.score due to commit:

commit 8c82d5a94b754b6015f8cf8ea1fde15821fc6611 ("x86 tsc_msr: Extend to include Intel Core Architecture")
https://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux.git release

in testcase: unixbench
on test machine: lkp-bdw-de1: 16 threads Broadwell-DE with 8G memory
with following parameters: cpufreq_governor=performance/nr_task=1/runtime=300s/test=whetstone-double



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/kconfig/nr_task/rootfs/runtime/tbox_group/test/testcase:
  gcc-4.9/performance/x86_64-rhel/1/debian-x86_64-2015-02-07.cgz/300s/lkp-bdw-de1/whetstone-double/unixbench

commit: 
  abc6a0f4b18281410da1a3f26e2819d8e03e144f
  8c82d5a94b754b6015f8cf8ea1fde15821fc6611

abc6a0f4b1828141 8c82d5a94b754b6015f8cf8ea1 
---------------- -------------------------- 
       fail:runs  %reproduction    fail:runs
           |             |             |    
          1:4          -25%            :4     kmsg.Spurious_LAPIC_timer_interrupt_on_cpu
         %stddev     %change         %stddev
             \          |                \  
    816.90 ±  0%     -15.6%     689.82 ±  0%  unixbench.score
    462.07 ±  0%      +2.3%     472.47 ±  0%  unixbench.time.elapsed_time
    462.07 ±  0%      +2.3%     472.47 ±  0%  unixbench.time.elapsed_time.max
     80.00 ±  0%      +1.2%      81.00 ±  0%  unixbench.time.percent_of_cpu_this_job_got
    371.83 ±  0%      +3.0%     383.12 ±  0%  unixbench.time.user_time
    613.50 ±  4%     +23.0%     754.50 ±  2%  vmstat.system.cs
     16034 ±  0%     +76.0%      28213 ±  1%  vmstat.system.in
      5852 ±  0%     +14.3%       6692 ±  4%  proc-vmstat.nr_active_anon
      2404 ±  0%     +24.5%       2993 ±  1%  proc-vmstat.nr_shmem
      1389 ±  1%     +62.4%       2257 ±  9%  proc-vmstat.pgactivate
     39227 ±  0%      +9.3%      42862 ±  2%  meminfo.Active
     23426 ±  0%     +14.3%      26777 ±  4%  meminfo.Active(anon)
      4082 ±  0%     +24.9%       5098 ± 19%  meminfo.AnonHugePages
      9619 ±  0%     +24.5%      11974 ±  1%  meminfo.Shmem
    498549 ±  4%    +367.4%    2330223 ±  3%  cpuidle.C1-BDW.time
     57855 ±  3%     +62.4%      93954 ±  2%  cpuidle.C1-BDW.usage
   6366987 ± 26%     -24.3%    4821327 ±  2%  cpuidle.C3-BDW.time
 6.651e+09 ±  0%     -72.3%  1.841e+09 ±  9%  cpuidle.C6-BDW.time
   7056683 ±  0%     -72.6%    1935634 ±  9%  cpuidle.C6-BDW.usage
  13526734 ± 34%  +37932.5%  5.145e+09 ±  3%  cpuidle.POLL.time
      1302 ±  2%   +3739.8%      50004 ±  6%  cpuidle.POLL.usage
      8.09 ±  0%    +826.0%      74.94 ±  3%  turbostat.%Busy
    189.50 ±  1%    +890.8%       1877 ±  3%  turbostat.Avg_MHz
     31.16 ±  0%     -75.6%       7.59 ±  9%  turbostat.CPU%c1
      0.04 ± 45%     -77.8%       0.01 ±  0%  turbostat.CPU%c3
     60.71 ±  0%     -71.2%      17.46 ± 10%  turbostat.CPU%c6
     56.25 ±  0%     +13.8%      64.00 ±  1%  turbostat.CoreTmp
      1.52 ±  2%     -99.7%       0.01 ±173%  turbostat.Pkg%pc2
      0.01 ± 34%    -100.0%       0.00 ± -1%  turbostat.Pkg%pc3
      9.95 ±  1%     -99.6%       0.04 ±116%  turbostat.Pkg%pc6
     55.25 ±  0%     +15.8%      64.00 ±  1%  turbostat.PkgTmp
     19.25 ±  0%     +55.8%      29.98 ±  1%  turbostat.PkgWatt
 1.396e+09 ±  0%     -43.3%   7.92e+08 ±  1%  perf-stat.L1-dcache-load-misses
 2.502e+11 ±  0%    +264.8%  9.128e+11 ±  2%  perf-stat.L1-dcache-loads
 1.205e+11 ±  1%     -19.8%  9.659e+10 ±  1%  perf-stat.L1-dcache-stores
 1.978e+09 ±  0%     -62.8%  7.362e+08 ±  7%  perf-stat.L1-icache-load-misses
  8.09e+08 ±  0%     -58.1%  3.391e+08 ±  4%  perf-stat.LLC-loads
 6.025e+08 ±  0%     -56.2%  2.641e+08 ±  4%  perf-stat.LLC-stores
  1.55e+11 ±  3%    +435.3%    8.3e+11 ±  3%  perf-stat.branch-instructions
 1.546e+09 ±  0%     -34.0%   1.02e+09 ±  1%  perf-stat.branch-load-misses
 1.519e+11 ±  0%    +445.3%  8.284e+11 ±  2%  perf-stat.branch-loads
 1.515e+09 ±  0%     -34.9%  9.861e+08 ±  1%  perf-stat.branch-misses
 5.222e+10 ±  0%    +971.5%  5.596e+11 ±  3%  perf-stat.bus-cycles
 4.109e+09 ±  0%     -62.0%  1.561e+09 ±  5%  perf-stat.cache-misses
 4.087e+09 ±  0%     -62.2%  1.547e+09 ±  6%  perf-stat.cache-references
    282484 ±  4%     +25.9%     355784 ±  2%  perf-stat.context-switches
 1.261e+12 ±  0%   +1010.8%  1.401e+13 ±  3%  perf-stat.cpu-cycles
 2.378e+08 ±  0%     -66.2%   80299694 ±  7%  perf-stat.dTLB-load-misses
 2.495e+11 ±  0%    +266.4%   9.14e+11 ±  2%  perf-stat.dTLB-loads
  39543763 ±  0%     -69.1%   12232186 ±  8%  perf-stat.dTLB-store-misses
 1.191e+11 ±  1%     -18.4%  9.722e+10 ±  0%  perf-stat.dTLB-stores
  89967876 ± 39%     -72.8%   24439509 ±  9%  perf-stat.iTLB-load-misses
  48505118 ±  0%     -73.9%   12652529 ±  5%  perf-stat.iTLB-loads
 1.341e+12 ±  3%    +193.5%  3.934e+12 ±  2%  perf-stat.instructions
    543009 ±  0%      +3.2%     560464 ±  1%  perf-stat.minor-faults
    542964 ±  0%      +3.2%     560437 ±  1%  perf-stat.page-faults
 1.097e+12 ±  0%    +971.6%  1.176e+13 ±  3%  perf-stat.ref-cycles
      1.97 ± 80%   +1290.8%      27.39 ± 84%  sched_debug.cfs_rq:/.MIN_vruntime.avg
     31.52 ± 80%   +1290.8%     438.31 ± 84%  sched_debug.cfs_rq:/.MIN_vruntime.max
      7.63 ± 80%   +1290.8%     106.10 ± 84%  sched_debug.cfs_rq:/.MIN_vruntime.stddev
    253.54 ±  6%      -9.6%     229.17 ±  6%  sched_debug.cfs_rq:/.load_avg.stddev
      1.97 ± 80%   +1292.0%      27.42 ± 84%  sched_debug.cfs_rq:/.max_vruntime.avg
     31.52 ± 80%   +1292.0%     438.68 ± 84%  sched_debug.cfs_rq:/.max_vruntime.max
      7.63 ± 80%   +1292.0%     106.19 ± 84%  sched_debug.cfs_rq:/.max_vruntime.stddev
    213.91 ±  5%      -9.5%     193.53 ±  0%  sched_debug.cpu.cpu_load[0].stddev
    901.94 ±  1%     -12.3%     790.97 ±  1%  sched_debug.cpu.cpu_load[1].max
    217.75 ±  2%     -12.5%     190.58 ±  0%  sched_debug.cpu.cpu_load[1].stddev
    895.38 ±  3%     -12.4%     784.59 ±  1%  sched_debug.cpu.cpu_load[2].max
    216.46 ±  4%     -12.7%     188.90 ±  1%  sched_debug.cpu.cpu_load[2].stddev
    880.84 ±  4%     -11.7%     777.62 ±  1%  sched_debug.cpu.cpu_load[3].max
    213.30 ±  5%     -12.3%     187.05 ±  1%  sched_debug.cpu.cpu_load[3].stddev
    867.22 ±  5%     -11.3%     769.41 ±  0%  sched_debug.cpu.cpu_load[4].max
    210.20 ±  5%     -11.8%     185.34 ±  0%  sched_debug.cpu.cpu_load[4].stddev
     10216 ±  2%     +18.4%      12096 ±  2%  sched_debug.cpu.nr_switches.avg
     28571 ± 16%     +48.8%      42525 ±  6%  sched_debug.cpu.nr_switches.max
      7768 ±  5%     +45.7%      11316 ±  8%  sched_debug.cpu.nr_switches.stddev
      8265 ±  2%     +21.7%      10056 ±  2%  sched_debug.cpu.sched_count.avg
     25938 ± 22%     +49.1%      38684 ± 10%  sched_debug.cpu.sched_count.max
      7071 ±  9%     +46.0%      10324 ± 10%  sched_debug.cpu.sched_count.stddev
      3783 ±  3%     +24.4%       4704 ±  2%  sched_debug.cpu.sched_goidle.avg
     11805 ± 16%     +56.1%      18427 ±  5%  sched_debug.cpu.sched_goidle.max
      3254 ±  5%     +55.0%       5045 ±  7%  sched_debug.cpu.sched_goidle.stddev
      3681 ±  3%     +24.4%       4580 ±  2%  sched_debug.cpu.ttwu_count.avg
    314.38 ±  3%     -27.9%     226.78 ±  9%  sched_debug.cpu.ttwu_local.min
      2.55 ± 26%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.__hrtimer_run_queues.hrtimer_interrupt.local_apic_timer_interrupt.smp_apic_timer_interrupt.apic_timer_interrupt
      1.94 ± 23%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.__tick_nohz_idle_enter.tick_nohz_irq_exit.irq_exit.smp_apic_timer_interrupt.apic_timer_interrupt
      9.41 ± 10%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.apic_timer_interrupt.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary
      1.40 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.call_console_drivers.constprop.23.console_unlock.vprintk_emit.vprintk_default.printk
      1.40 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.console_unlock.vprintk_emit.vprintk_default.printk.perf_duration_warn
      2.31 ± 10%    +162.8%       6.08 ± 52%  perf-profile.cycles-pp.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry.rest_init
     81.21 ±  1%     +15.0%      93.40 ±  3%  perf-profile.cycles-pp.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry.start_secondary
      1.12 ± 17%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.cpuidle_select.cpu_startup_entry.start_secondary
      3.33 ± 21%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.hrtimer_interrupt.local_apic_timer_interrupt.smp_apic_timer_interrupt.apic_timer_interrupt.cpuidle_enter
     81.30 ±  1%     -98.7%       1.07 ± 16%  perf-profile.cycles-pp.intel_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry
      0.92 ± 16%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.io_serial_in.wait_for_xmitr.serial8250_console_putchar.uart_console_write.serial8250_console_write
      1.98 ±  8%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.irq_enter.smp_apic_timer_interrupt.apic_timer_interrupt.cpuidle_enter.call_cpuidle
      2.65 ± 11%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.irq_exit.smp_apic_timer_interrupt.apic_timer_interrupt.cpuidle_enter.call_cpuidle
      1.40 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.irq_work_run.smp_irq_work_interrupt.irq_work_interrupt.cpuidle_enter.call_cpuidle
      1.40 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.irq_work_run_list.irq_work_run.smp_irq_work_interrupt.irq_work_interrupt.cpuidle_enter
      3.52 ± 21%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.local_apic_timer_interrupt.smp_apic_timer_interrupt.apic_timer_interrupt.cpuidle_enter.call_cpuidle
      1.40 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.perf_duration_warn.irq_work_run_list.irq_work_run.smp_irq_work_interrupt.irq_work_interrupt
      1.02 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.perf_mux_hrtimer_handler.__hrtimer_run_queues.hrtimer_interrupt.local_apic_timer_interrupt.smp_apic_timer_interrupt
      0.54 ±173%  +18227.6%      98.05 ±  0%  perf-profile.cycles-pp.poll_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.cpu_startup_entry
      1.40 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.printk.perf_duration_warn.irq_work_run_list.irq_work_run.smp_irq_work_interrupt
      1.29 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.serial8250_console_putchar.uart_console_write.serial8250_console_write.univ8250_console_write.call_console_drivers.constprop.23
      1.29 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.serial8250_console_write.univ8250_console_write.call_console_drivers.constprop.23.console_unlock.vprintk_emit
      9.54 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.smp_apic_timer_interrupt.apic_timer_interrupt.cpuidle_enter.call_cpuidle.cpu_startup_entry
      1.40 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.smp_irq_work_interrupt.irq_work_interrupt.cpuidle_enter.call_cpuidle.cpu_startup_entry
      1.24 ± 26%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.tick_irq_enter.irq_enter.smp_apic_timer_interrupt.apic_timer_interrupt.cpuidle_enter
      2.03 ± 22%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.tick_nohz_irq_exit.irq_exit.smp_apic_timer_interrupt.apic_timer_interrupt.cpuidle_enter
      1.50 ± 25%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.tick_nohz_stop_sched_tick.__tick_nohz_idle_enter.tick_nohz_irq_exit.irq_exit.smp_apic_timer_interrupt
      1.29 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.uart_console_write.serial8250_console_write.univ8250_console_write.call_console_drivers.constprop.23.console_unlock
      1.29 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.univ8250_console_write.call_console_drivers.constprop.23.console_unlock.vprintk_emit.vprintk_default
      1.40 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.vprintk_default.printk.perf_duration_warn.irq_work_run_list.irq_work_run
      1.40 ± 12%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.vprintk_emit.vprintk_default.printk.perf_duration_warn.irq_work_run_list
      1.26 ± 14%    -100.0%       0.00 ± -1%  perf-profile.cycles-pp.wait_for_xmitr.serial8250_console_putchar.uart_console_write.serial8250_console_write.univ8250_console_write





                                  unixbench.score

  820 *+*--*-*-*--*-*-*--*-*-*--*-*-*--*-*-*--*-*-*--*-*-*--*-*-*--*--------+
      |                                                                     |
  800 ++                                                                    |
      |                                                                     |
  780 ++                                                                    |
      |                                                                     |
  760 ++                                                                    |
      |                                                                     |
  740 ++                                                                    |
      |                                                                     |
  720 ++                                                                    |
      |                                                                     |
  700 ++                                                                    |
      O O  O O O  O O O  O O O  O O O  O O O  O O O  O O O  O O O  O O O  O O
  680 ++--------------------------------------------------------------------+


	[*] 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

Attachments

Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help