[LKP] [mm] 10359213d05: -3.1% pigz.time.minor_page_faults, +31.3% meminfo.AnonHugePages, +1.7% will-it-scale.per_thread_ops
From: Huang Ying
Date: Mon Feb 16 2015 - 00:11:42 EST
FYI, we noticed the below changes on
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master
commit 10359213d05acf804558bda7cc9b8422a828d1cd ("mm: incorporate read-only pages into transparent huge pages")
testbox/testcase/testparams: xps2/pigz/performance-100%-512K
ba4877b9ca51f80b 10359213d05acf804558bda7cc
---------------- --------------------------
fail:runs %reproduction fail:runs
| | |
%stddev %change %stddev
\ | \
25068 Â 3% +8.1% 27104 Â 3% pigz.time.maximum_resident_set_size
27037 Â 0% -3.1% 26197 Â 1% pigz.time.minor_page_faults
12558 Â 2% +31.3% 16493 Â 3% meminfo.AnonHugePages
0.92 Â 30% -45.1% 0.51 Â 25% perf-profile.cpu-cycles.update_curr.task_tick_fair.scheduler_tick.update_process_times.tick_sched_handle
1.12 Â 3% +42.1% 1.59 Â 15% perf-profile.cpu-cycles.ret_from_fork
2.46 Â 21% -29.4% 1.74 Â 18% perf-profile.cpu-cycles.get_page_from_freelist.__alloc_pages_nodemask.alloc_pages_current.pipe_write.new_sync_write
595245 Â 11% -26.6% 436928 Â 23% sched_debug.cpu#3.avg_idle
602523 Â 9% -25.1% 451132 Â 15% sched_debug.cpu#5.avg_idle
16.26 Â 9% -16.0% 13.66 Â 4% perf-profile.cpu-cycles.tick_program_event.hrtimer_interrupt.hpet_interrupt_handler.handle_irq_event_percpu.handle_irq_event
0.81 Â 24% +30.2% 1.05 Â 5% perf-profile.cpu-cycles.__do_softirq.irq_exit.smp_apic_timer_interrupt.apic_timer_interrupt
1.68 Â 5% -22.2% 1.30 Â 22% perf-profile.cpu-cycles.copy_page_from_iter_iovec.copy_page_from_iter.pipe_write.new_sync_write.vfs_write
150 Â 2% -11.8% 132 Â 8% sched_debug.cfs_rq[3]:/.load
7785 Â 6% +11.4% 8675 Â 6% slabinfo.kmalloc-64.active_objs
1986 Â 3% -12.4% 1739 Â 4% slabinfo.kmalloc-256.active_objs
8048 Â 4% +9.7% 8826 Â 5% slabinfo.kmalloc-64.num_objs
321 Â 28% +61.6% 520 Â 28% sched_debug.cfs_rq[2]:/.blocked_load_avg
139 Â 3% +14.0% 159 Â 7% sched_debug.cpu#1.load
121 Â 2% +8.2% 131 Â 3% sched_debug.cpu#0.cpu_load[4]
122 Â 2% +9.2% 134 Â 3% sched_debug.cpu#0.cpu_load[3]
1014549 Â 8% -9.8% 914927 Â 4% cpuidle.C1E-NHM.time
25068 Â 3% +8.1% 27104 Â 3% time.maximum_resident_set_size
143231 Â 9% +11.4% 159600 Â 2% softirqs.RCU
150 Â 2% -12.1% 132 Â 8% sched_debug.cpu#3.load
testbox/testcase/testparams: lkp-sbx04/will-it-scale/performance-malloc1
ba4877b9ca51f80b 10359213d05acf804558bda7cc
---------------- --------------------------
3954711 Â 4% +11.8% 4420591 Â 8% will-it-scale.time.voluntary_context_switches
7468699 Â 1% +3.4% 7722740 Â 1% will-it-scale.time.minor_page_faults
0.02 Â 0% -3.7% 0.02 Â 0% will-it-scale.scalability
37784 Â 0% +1.7% 38421 Â 0% will-it-scale.per_thread_ops
0.00 Â 0% +5.8e+12% 58403.14 Â 29% sched_debug.cfs_rq[49]:/.MIN_vruntime
0.00 Â 0% +5.8e+12% 58403.14 Â 29% sched_debug.cfs_rq[49]:/.max_vruntime
5 Â 24% +340.0% 22 Â 13% sched_debug.cpu#59.cpu_load[1]
4 Â 17% +310.5% 19 Â 17% sched_debug.cpu#59.cpu_load[0]
34 Â 29% +198.5% 101 Â 31% sched_debug.cfs_rq[33]:/.tg_load_contrib
5 Â 24% +280.0% 19 Â 15% sched_debug.cpu#59.cpu_load[2]
19 Â 41% -79.5% 4 Â 50% sched_debug.cfs_rq[26]:/.load
19 Â 41% -79.5% 4 Â 50% sched_debug.cpu#26.load
33117 Â 39% +159.0% 85783 Â 15% sched_debug.cpu#4.sched_goidle
4 Â 19% +211.1% 14 Â 14% sched_debug.cpu#59.cpu_load[3]
34216 Â 41% +175.2% 94151 Â 18% sched_debug.cpu#4.ttwu_count
7 Â 34% +160.0% 19 Â 17% sched_debug.cfs_rq[49]:/.load
124 Â 21% -78.8% 26 Â 43% sched_debug.cfs_rq[48]:/.blocked_load_avg
66389 Â 39% +158.7% 171750 Â 15% sched_debug.cpu#4.nr_switches
135 Â 23% -74.1% 35 Â 28% sched_debug.cfs_rq[48]:/.tg_load_contrib
0.00 Â 1% +166.2% 0.00 Â 34% sched_debug.rt_rq[63]:/.rt_time
5 Â 47% -63.6% 2 Â 0% sched_debug.cpu#11.cpu_load[2]
67305 Â 38% +155.4% 171876 Â 15% sched_debug.cpu#4.sched_count
47 Â 40% -68.3% 15 Â 20% sched_debug.cfs_rq[12]:/.tg_load_contrib
54 Â 20% +149.8% 135 Â 44% sched_debug.cfs_rq[45]:/.blocked_load_avg
0.00 Â 7% -15.3% 0.00 Â 4% sched_debug.rt_rq[7]:/.rt_time
5 Â 37% -65.2% 2 Â 0% sched_debug.cpu#3.cpu_load[1]
1 Â 34% +100.0% 2 Â 20% sched_debug.cfs_rq[19]:/.nr_spread_over
136 Â 42% -62.3% 51 Â 8% sched_debug.cfs_rq[43]:/.tg_load_contrib
67 Â 16% +124.6% 150 Â 41% sched_debug.cfs_rq[45]:/.tg_load_contrib
69 Â 36% +126.1% 156 Â 21% sched_debug.cfs_rq[61]:/.blocked_load_avg
4 Â 10% +135.3% 10 Â 10% sched_debug.cpu#59.cpu_load[4]
7 Â 14% -74.2% 2 Â 50% sched_debug.cpu#49.nr_uninterruptible
877 Â 32% +64.3% 1442 Â 30% sched_debug.cpu#22.curr->pid
7 Â 29% +151.6% 19 Â 28% sched_debug.cpu#50.load
4 Â 22% -57.9% 2 Â 0% sched_debug.cpu#3.cpu_load[2]
347606 Â 36% -58.7% 143557 Â 1% sched_debug.cfs_rq[16]:/.min_vruntime
0.00 Â 39% -47.9% 0.00 Â 0% sched_debug.rt_rq[55]:/.rt_time
7757 Â 45% -61.5% 2985 Â 5% sched_debug.cfs_rq[16]:/.exec_clock
15 Â 36% -42.9% 9 Â 0% sched_debug.cpu#0.cpu_load[0]
77 Â 32% +120.7% 170 Â 16% sched_debug.cfs_rq[61]:/.tg_load_contrib
8767 Â 9% +70.0% 14901 Â 33% sched_debug.cfs_rq[1]:/.exec_clock
406 Â 36% +35.5% 550 Â 29% sched_debug.cpu#46.ttwu_local
97714 Â 37% +80.7% 176528 Â 19% sched_debug.cpu#6.nr_switches
98697 Â 36% +79.0% 176658 Â 19% sched_debug.cpu#6.sched_count
88 Â 14% -60.0% 35 Â 23% sched_debug.cfs_rq[62]:/.blocked_load_avg
0.00 Â 36% -48.5% 0.00 Â 10% sched_debug.rt_rq[54]:/.rt_time
1 Â 0% +100.0% 2 Â 0% sched_debug.cfs_rq[53]:/.nr_spread_over
3 Â 11% -46.7% 2 Â 0% sched_debug.cpu#3.cpu_load[3]
5714 Â 27% -37.4% 3577 Â 4% sched_debug.cpu#28.sched_goidle
745 Â 26% +48.1% 1104 Â 20% sched_debug.cpu#15.ttwu_local
11686 Â 26% -37.0% 7366 Â 4% sched_debug.cpu#28.nr_switches
90749 Â 11% +54.3% 140005 Â 29% sched_debug.cpu#32.sched_goidle
181994 Â 11% +54.1% 280377 Â 29% sched_debug.cpu#32.nr_switches
182550 Â 10% +53.8% 280771 Â 29% sched_debug.cpu#32.sched_count
48790 Â 37% +80.7% 88182 Â 19% sched_debug.cpu#6.sched_goidle
48101669 Â 16% +58.7% 76337780 Â 27% cpuidle.C1-SNB.time
11390 Â 16% +62.0% 18455 Â 28% sched_debug.cpu#11.ttwu_count
50734 Â 38% +79.4% 91038 Â 16% sched_debug.cpu#6.ttwu_count
15 Â 21% -34.4% 10 Â 0% sched_debug.cpu#0.cpu_load[1]
63 Â 39% +96.1% 125 Â 12% sched_debug.cfs_rq[58]:/.blocked_load_avg
409 Â 9% +47.9% 605 Â 26% sched_debug.cpu#59.ttwu_local
1267 Â 23% +22.4% 1552 Â 25% numa-meminfo.node3.AnonHugePages
72 Â 25% +92.4% 140 Â 3% sched_debug.cfs_rq[58]:/.tg_load_contrib
0.00 Â 32% -43.6% 0.00 Â 1% sched_debug.rt_rq[32]:/.rt_time
1147 Â 24% +44.2% 1654 Â 10% sched_debug.cpu#25.ttwu_local
5 Â 14% -60.0% 2 Â 50% sched_debug.cpu#51.nr_uninterruptible
12 Â 42% +95.9% 24 Â 4% sched_debug.cfs_rq[50]:/.load
14738 Â 6% +53.2% 22584 Â 7% sched_debug.cpu#12.sched_goidle
29667 Â 6% +52.8% 45321 Â 7% sched_debug.cpu#12.nr_switches
14586 Â 8% +54.0% 22463 Â 7% sched_debug.cpu#12.ttwu_count
1025 Â 17% -41.3% 601 Â 10% sched_debug.cpu#12.ttwu_local
26504 Â 33% +57.3% 41688 Â 19% sched_debug.cpu#9.sched_count
73 Â 38% +68.7% 124 Â 3% sched_debug.cfs_rq[29]:/.tg_load_contrib
707 Â 20% +63.1% 1154 Â 8% numa-meminfo.node3.Shmem
94 Â 12% -50.0% 47 Â 34% sched_debug.cfs_rq[62]:/.tg_load_contrib
176 Â 20% +63.2% 288 Â 9% numa-vmstat.node3.nr_shmem
1378 Â 15% +15.9% 1597 Â 21% sched_debug.cpu#1.curr->pid
69 Â 48% +77.5% 122 Â 2% sched_debug.cfs_rq[29]:/.blocked_load_avg
692 Â 2% +25.0% 865 Â 21% sched_debug.cpu#30.curr->pid
17 Â 33% +52.1% 27 Â 18% sched_debug.cpu#46.load
3 Â 0% -33.3% 2 Â 0% sched_debug.cpu#3.cpu_load[4]
390329 Â 0% +24.1% 484415 Â 19% sched_debug.cfs_rq[1]:/.min_vruntime
635 Â 20% +65.9% 1054 Â 5% numa-meminfo.node3.Inactive(anon)
158 Â 20% +66.2% 263 Â 5% numa-vmstat.node3.nr_inactive_anon
140 Â 4% +26.7% 178 Â 16% sched_debug.cfs_rq[1]:/.tg_runnable_contrib
6553 Â 19% -31.8% 4467 Â 17% sched_debug.cpu#26.sched_goidle
6476 Â 4% +26.5% 8195 Â 16% sched_debug.cfs_rq[1]:/.avg->runnable_avg_sum
719 Â 31% +46.9% 1056 Â 3% sched_debug.cpu#16.ttwu_local
278733 Â 11% +28.3% 357584 Â 11% cpuidle.C1E-SNB.usage
12547 Â 10% +35.0% 16933 Â 6% sched_debug.cpu#58.sched_count
6341 Â 10% +35.5% 8592 Â 2% sched_debug.cpu#58.ttwu_count
139557 Â 14% +18.4% 165256 Â 10% sched_debug.cfs_rq[59]:/.spread0
1767 Â 3% +29.4% 2286 Â 3% sched_debug.cpu#61.curr->pid
136520 Â 13% +16.7% 159351 Â 9% sched_debug.cfs_rq[63]:/.spread0
142974 Â 4% +16.3% 166215 Â 14% sched_debug.cfs_rq[25]:/.min_vruntime
13231 Â 23% +29.8% 17174 Â 12% sched_debug.cpu#16.sched_count
113 Â 35% -34.5% 74 Â 0% sched_debug.cfs_rq[25]:/.tg_load_contrib
23 Â 19% +38.3% 32 Â 10% sched_debug.cfs_rq[37]:/.load
1359 Â 21% -40.1% 814 Â 16% sched_debug.cpu#16.curr->pid
141008 Â 16% +16.6% 164377 Â 9% sched_debug.cfs_rq[61]:/.spread0
107 Â 32% -31.8% 73 Â 1% sched_debug.cfs_rq[25]:/.blocked_load_avg
39693 Â 7% -20.7% 31477 Â 0% sched_debug.cpu#44.ttwu_count
960 Â 14% +38.2% 1327 Â 1% sched_debug.cpu#22.ttwu_local
827 Â 11% +20.2% 995 Â 9% sched_debug.cpu#23.curr->pid
1819 Â 5% +18.8% 2162 Â 7% sched_debug.cpu#60.curr->pid
6671 Â 13% +24.4% 8297 Â 8% sched_debug.cpu#60.sched_goidle
13517 Â 13% +24.2% 16793 Â 8% sched_debug.cpu#60.nr_switches
13658 Â 12% +24.0% 16934 Â 8% sched_debug.cpu#60.sched_count
39297 Â 8% -19.8% 31521 Â 1% sched_debug.cpu#44.sched_goidle
835 Â 13% +30.6% 1090 Â 1% sched_debug.cpu#18.ttwu_local
78817 Â 8% -19.8% 63241 Â 1% sched_debug.cpu#44.nr_switches
79478 Â 8% -20.0% 63595 Â 1% sched_debug.cpu#44.sched_count
14 Â 10% -25.4% 11 Â 9% sched_debug.cpu#0.cpu_load[2]
1.43 Â 1% -20.6% 1.14 Â 2% perf-profile.cpu-cycles.up_write.vm_munmap.sys_munmap.system_call_fastpath
6092 Â 10% +28.1% 7803 Â 0% sched_debug.cpu#58.sched_goidle
348 Â 13% +18.7% 413 Â 10% sched_debug.cpu#6.ttwu_local
12410 Â 10% +27.3% 15796 Â 0% sched_debug.cpu#58.nr_switches
7012 Â 15% +25.7% 8813 Â 7% sched_debug.cpu#60.ttwu_count
377 Â 11% +13.5% 428 Â 7% sched_debug.cpu#60.ttwu_local
9050 Â 6% +15.8% 10476 Â 1% sched_debug.cfs_rq[4]:/.exec_clock
1762 Â 2% +8.1% 1906 Â 8% sched_debug.cpu#62.curr->pid
3954711 Â 4% +11.8% 4420591 Â 8% time.voluntary_context_switches
2045 Â 5% +11.9% 2290 Â 7% numa-meminfo.node3.KernelStack
14 Â 8% -14.3% 12 Â 0% sched_debug.cpu#43.cpu_load[4]
371 Â 1% -17.0% 308 Â 5% sched_debug.cpu#44.ttwu_local
15 Â 15% -20.6% 12 Â 4% sched_debug.cpu#43.cpu_load[3]
23 Â 7% -14.0% 20 Â 0% sched_debug.cpu#34.cpu_load[3]
0.00 Â 4% -22.2% 0.00 Â 14% sched_debug.rt_rq[50]:/.rt_time
374746 Â 13% +18.8% 445099 Â 3% sched_debug.cfs_rq[50]:/.spread0
650306 Â 5% +13.4% 737674 Â 2% sched_debug.cpu#33.avg_idle
2796 Â 5% +3.9% 2905 Â 6% sched_debug.cpu#36.curr->pid
388 Â 7% +14.6% 444 Â 3% sched_debug.cpu#63.ttwu_local
641680 Â 2% +9.2% 700581 Â 5% sched_debug.cpu#39.avg_idle
9736 Â 8% +13.1% 11014 Â 2% sched_debug.cfs_rq[6]:/.exec_clock
3356 Â 2% +8.6% 3646 Â 3% sched_debug.cfs_rq[29]:/.avg->runnable_avg_sum
71 Â 3% +8.7% 78 Â 3% sched_debug.cfs_rq[29]:/.tg_runnable_contrib
16875 Â 3% +9.5% 18477 Â 2% slabinfo.vm_area_struct.active_objs
16875 Â 3% +9.5% 18477 Â 2% slabinfo.vm_area_struct.num_objs
1.17 Â 4% -9.0% 1.06 Â 1% perf-profile.cpu-cycles.perf_event_mmap.mmap_region.do_mmap_pgoff.vm_mmap_pgoff.sys_mmap_pgoff
44305 Â 4% -16.1% 37164 Â 8% sched_debug.cpu#43.ttwu_count
6430 Â 1% +6.4% 6840 Â 5% sched_debug.cfs_rq[14]:/.avg->runnable_avg_sum
2768 Â 3% +3.7% 2870 Â 5% sched_debug.cpu#38.curr->pid
9224 Â 2% +9.8% 10130 Â 1% slabinfo.anon_vma.num_objs
9224 Â 2% +9.8% 10130 Â 1% slabinfo.anon_vma.active_objs
1816 Â 4% +10.1% 2000 Â 2% sched_debug.cpu#57.curr->pid
85823 Â 4% -15.5% 72483 Â 7% sched_debug.cpu#43.nr_switches
86191 Â 4% -15.5% 72848 Â 7% sched_debug.cpu#43.sched_count
44120 Â 12% +18.1% 52091 Â 4% sched_debug.cpu#1.nr_load_updates
42455 Â 5% -14.9% 36139 Â 7% sched_debug.cpu#43.sched_goidle
27096 Â 3% +10.6% 29973 Â 8% vmstat.system.cs
18715 Â 0% +7.0% 20016 Â 5% vmstat.system.in
testbox/testcase/testparams: snb-drag/phoronix-test-suite/performance-encode-ogg-1.4.1
ba4877b9ca51f80b 10359213d05acf804558bda7cc
---------------- --------------------------
29266 Â 0% -13.0% 25474 Â 1% phoronix-test-suite.time.minor_page_faults
320 Â 35% +367.6% 1496 Â 44% sched_debug.cfs_rq[2]:/.tg_load_contrib
63 Â 35% +103.1% 129 Â 28% sched_debug.cpu#2.cpu_load[1]
68 Â 37% +148.7% 171 Â 26% sched_debug.cpu#2.cpu_load[0]
126 Â 37% +73.5% 219 Â 38% sched_debug.cfs_rq[2]:/.runnable_load_avg
298 Â 3% -25.5% 222 Â 19% sched_debug.cfs_rq[3]:/.load
35 Â 8% +21.8% 43 Â 10% sched_debug.cfs_rq[1]:/.nr_spread_over
29266 Â 0% -13.0% 25474 Â 1% time.minor_page_faults
xps2: Nehalem
Memory: 4G
lkp-sbx04: Sandy Bridge-EX
Memory: 64G
snb-drag: Sandy Bridge
Memory: 6G
pigz.time.minor_page_faults
27400 ++------------------------------------------------------------------+
27200 ++ * ** |
| .* *** *.* ** +: .** .** + : .* *.* *. *
27000 ** ***.***.** + + ** * .* * :** * ** *.** * * ** *|
26800 ++ ** * * * * |
| |
26600 ++ O |
26400 ++ O |
26200 ++ O O O O O O O O O OO O |
O |
26000 +O O O O O OO O O |
25800 ++ O O O OO O |
| |
25600 ++ O O |
25400 ++---------------------------------O-O------------------------------+
[*] bisect-good sample
[O] bisect-bad sample
To reproduce:
apt-get install ruby
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,
Ying Huang
---
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: ea2bbe3b9bf930408db205344fe10c8f719ba738
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: 2015-02-15 01:37:17.922838823 +08:00
head_commit: ea2bbe3b9bf930408db205344fe10c8f719ba738
base_commit: bfa76d49576599a4b9f9b7a71f23d73d6dcff735
branch: linux-devel/devel-hourly-2015021423
kernel: "/kernel/x86_64-rhel/ea2bbe3b9bf930408db205344fe10c8f719ba738/vmlinuz-3.19.0-gea2bbe3"
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/ea2bbe3b9bf930408db205344fe10c8f719ba738/0"
job_file: "/lkp/scheduled/xps2/cyclic_pigz-performance-100%-512K-x86_64-rhel-HEAD-ea2bbe3b9bf930408db205344fe10c8f719ba738-0-20150215-58738-yzfujf.yaml"
dequeue_time: 2015-02-15 01:49:47.299361589 +08:00
nr_cpu: "$(nproc)"
job_state: finished
loadavg: 6.66 5.14 2.31 1/132 5309
start_time: '1423936212'
end_time: '1423936513'
version: "/lkp/lkp/.src-20150213-094846"
_______________________________________________
LKP mailing list
LKP@xxxxxxxxxxxxxxx