Skip to contentSkip to conversation

7d5b506f3c2: -3.9% pigz.time.minor_page_faults

LKP Huang Ying 2 participants 2 messages Feb 12, 2015

lore
Share & tools
lore

Huang Ying

author
#0 lore raw

FYI, we noticed the below changes on

git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
commit 7d5b506f3c295ff81e34235544a9e6ac143743c1 ("mm: incorporate read-only pages into transparent huge pages")

testbox/testcase/testparams: xps2/pigz/performance-100%-512K

c3916839649ddb3c 7d5b506f3c295ff81e34235544
---------------- --------------------------

         %stddev     %change         %stddev
             \          |                \  
     27105 ±  0%      -3.9%      26056 ±  1%  pigz.time.minor_page_faults
      0.93 ± 49%     -55.0%       0.42 ± 17%  perf-profile.cpu-cycles.enqueue_task.activate_task.ttwu_do_activate.try_to_wake_up.default_wake_function
      0.83 ± 43%     -49.5%       0.42 ±  9%  perf-profile.cpu-cycles.try_to_wake_up.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key
      2.53 ± 37%     -41.0%       1.49 ±  6%  perf-profile.cpu-cycles.__alloc_pages_nodemask.alloc_pages_current.pipe_write.new_sync_write.vfs_write
      0.77 ± 47%     -49.0%       0.40 ±  2%  perf-profile.cpu-cycles.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.pipe_read.new_sync_read
      0.77 ± 47%     -49.4%       0.39 ±  3%  perf-profile.cpu-cycles.default_wake_function.autoremove_wake_function.__wake_up_common.__wake_up_sync_key.pipe_read
      0.77 ± 47%     -46.1%       0.42 ± 10%  perf-profile.cpu-cycles.__wake_up_common.__wake_up_sync_key.pipe_read.new_sync_read.__vfs_read
      1.62 ± 37%     -40.8%       0.96 ± 14%  perf-profile.cpu-cycles.get_page_from_freelist.__alloc_pages_nodemask.alloc_pages_current.pipe_write.new_sync_write
      0.78 ± 18%     -47.7%       0.40 ± 16%  perf-profile.cpu-cycles.run_timer_softirq.__do_softirq.irq_exit.do_IRQ.ret_from_intr
      2.78 ± 32%     -35.1%       1.80 ±  8%  perf-profile.cpu-cycles.alloc_pages_current.pipe_write.new_sync_write.vfs_write.sys_write
      1.09 ± 28%     -35.5%       0.71 ± 23%  perf-profile.cpu-cycles.put_page.anon_pipe_buf_release.pipe_read.new_sync_read.__vfs_read
      1.48 ± 15%     -33.5%       0.98 ± 17%  perf-profile.cpu-cycles.__do_softirq.irq_exit.do_IRQ.ret_from_intr
      1.63 ± 20%     -30.9%       1.12 ± 14%  perf-profile.cpu-cycles.anon_pipe_buf_release.pipe_read.new_sync_read.__vfs_read.vfs_read
     15.76 ± 10%     +27.5%      20.09 ±  6%  perf-profile.cpu-cycles.tick_program_event.hrtimer_interrupt.hpet_interrupt_handler.handle_irq_event_percpu.handle_irq_event
     15.61 ± 10%     +26.8%      19.79 ±  6%  perf-profile.cpu-cycles.clockevents_program_event.tick_program_event.hrtimer_interrupt.hpet_interrupt_handler.handle_irq_event_percpu
     15.21 ±  9%     +27.9%      19.45 ±  6%  perf-profile.cpu-cycles.hpet_msi_next_event.clockevents_program_event.tick_program_event.hrtimer_interrupt.hpet_interrupt_handler
      1758 ± 36%     +46.6%       2578 ±  3%  sched_debug.cpu#0.curr->pid
      2.43 ± 10%     +27.9%       3.11 ±  3%  perf-profile.cpu-cycles.irq_exit.smp_apic_timer_interrupt.apic_timer_interrupt
       130 ±  4%     +13.8%        148 ±  5%  sched_debug.cpu#3.load
       130 ±  4%     +13.8%        148 ±  5%  sched_debug.cfs_rq[3]:/.load
      2498 ±  5%     +10.4%       2757 ±  6%  slabinfo.kmalloc-256.num_objs

xps2: Nehalem
Memory: 4G

pigz.time.minor_page_faults

27400 ++------------------------------------------------------------------+
27200 ++ *. *.* .* |
| .*. + * + + * * *.* + .*

  27000 *+*   *.*.*.*.*.**    +   .*.*.*.*.*.*.*.*. .*     * : : + +      * |
  26800 ++                     *.*                 *         : :  *         |
        |                                                     *             |
  26600 ++                                                                  |
  26400 ++                       O             O                            |
  26200 ++                         O       O                                |
        |           O O  O                                                  |
  26000 ++O                                                                 |
  25800 ++  O O         O  O O O       O O   O                              |
        |       O O                                                         |
  25600 O+                           O                                      |

25400 ++------------------------------------------------------------------+

[*] bisect-good sample
[O] bisect-bad sample

To reproduce:

apt-get install ruby ruby-oj
git clone git://git.kernel.org/pub/scm/linux/kernel/git/wfg/lkp-tests.git
cd lkp-tests
bin/setup-local job.yaml # the job file attached in this email
bin/run-local 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,
Huang, Ying

_______________________________________________
LKP mailing list
LKP(a)linux.intel.com

---
testcase: pigz
default-monitors:
wait: pre-test
uptime:
iostat:
vmstat:
numa-numastat:
numa-vmstat:
numa-meminfo:
proc-vmstat:
proc-stat:
meminfo:
slabinfo:
interrupts:
lock_stat:
latency_stats:
softirqs:
bdi_dev_mapping:
diskstats:
nfsstat:
cpuidle:
cpufreq-stats:
turbostat:
pmeter:
sched_debug:
interval: 10
default_watchdogs:
watch-oom:
watchdog:
cpufreq_governor: performance
commit: f61b8d6d777dfb241ebe87478c5c82126ac8d687
model: Nehalem
memory: 4G
nr_ssd_partitions: 1
ssd_partitions: "/dev/disk/by-id/ata-INTEL_SSDSC2CW240A3_CVCV204303XB240CGN-part1"
swap_partitions:
rootfs_partition: "/dev/disk/by-id/ata-INTEL_SSDSC2CW240A3_CVCV204303XB240CGN-part2"
netconsole_port: 6665
perf-profile:
freq: 800
nr_threads: 100%
pigz:
blocksize: 512K
testbox: xps2
tbox_group: xps2
kconfig: x86_64-rhel
enqueue_time: 2015-02-11 08:54:24.084863795 +08:00
head_commit: f61b8d6d777dfb241ebe87478c5c82126ac8d687
base_commit: bfa76d49576599a4b9f9b7a71f23d73d6dcff735
branch: next/master
kernel: "/kernel/x86_64-rhel/f61b8d6d777dfb241ebe87478c5c82126ac8d687/vmlinuz-3.19.0-next-20150210-gf61b8d6"
user: lkp
queue: cyclic
rootfs: debian-x86_64-2015-02-07.cgz
result_root: "/result/xps2/pigz/performance-100%-512K/debian-x86_64-2015-02-07.cgz/x86_64-rhel/f61b8d6d777dfb241ebe87478c5c82126ac8d687/0"
job_file: "/lkp/scheduled/xps2/cyclic_pigz-performance-100%-512K-x86_64-rhel-HEAD-f61b8d6d777dfb241ebe87478c5c82126ac8d687-0-20150211-14981-iea6fd.yaml"
dequeue_time: 2015-02-11 09:08:52.086449889 +08:00
nr_cpu: "$(nproc)"
job_state: finished
loadavg: 6.38 5.04 2.28 1/133 5308
start_time: '1423616958'
end_time: '1423617259'
version: "/lkp/lkp/.src-20150210-213435"

#1 lore raw

Hi Huang,

Quoted 5 lines · On Thu, 12 Feb 2015 09:35:26 +0800 Huang Ying <[email protected]> wrote:

FYI, we noticed the below changes on

git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
commit 7d5b506f3c295ff81e34235544a9e6ac143743c1 ("mm: incorporate read-only pages into transparent huge pages")

You may need to be more precise about the tree that you are using as linux-next rebases every day and so there could be different interactions with a single commit depending on what else is in a particular release. For example, the above commit appears in 4 different linux-next releases so far (next-20150204, next-20150209, next-20150210, next-20150211).

Also, mm related problems should probably also be reported to Andrew Morton.

Signature · 2 lines
Cheers,
Stephen Rothwell                    sfr(a)canb.auug.org.au