[LKP] [regression] d7aa432ad5f: -5.6% will-it-scale.per_thread_ops

From: Huang Ying
Date: Wed Mar 18 2015 - 04:08:43 EST


FYI, we noticed the below changes on

git://flatbed.openfabrics.org/~amirv/linux.git for-upstream
commit d7aa432ad5fa589c56a11bd7993cff05e9b87d05 ("regression: Disable FENTRY [topic: regression] [net-next:regression]")


testbox/testcase/testparams: lkp-sb03/will-it-scale/poll2

5cd08c4aa1751f04 d7aa432ad5fa589c56a11bd799
---------------- --------------------------
%stddev %change %stddev
\ | \
650868 Â 0% -2.9% 631737 Â 0% will-it-scale.per_process_ops
281122 Â 0% -5.6% 265461 Â 0% will-it-scale.per_thread_ops
153 Â 0% -7.9% 141 Â 0% will-it-scale.time.user_time
118 Â 0% +0.9% 119 Â 0% turbostat.CorWatt
2733 Â 3% +12.2% 3066 Â 2% proc-vmstat.nr_alloc_batch
1319 Â 3% +19.1% 1570 Â 5% numa-vmstat.node0.nr_alloc_batch
2250 Â 6% -13.0% 1958 Â 5% slabinfo.kmalloc-2048.active_objs
2380 Â 6% -13.2% 2066 Â 5% slabinfo.kmalloc-2048.num_objs
31.65 Â 0% -92.2% 2.46 Â 7% perf-profile.cpu-cycles.fput.sys_poll.system_call_fastpath
60.25 Â 0% +46.7% 88.37 Â 5% perf-profile.cpu-cycles.do_sys_poll.sys_poll.system_call_fastpath
3.57 Â 7% -10.2% 3.21 Â 3% perf-profile.cpu-cycles.copy_user_generic_string.sys_poll.system_call_fastpath
2.66 Â 2% -100.0% 0.00 Â 0% perf-profile.cpu-cycles.__fdget.sys_poll.system_call_fastpath
2.71 Â 0% -100.0% 0.00 Â 0% perf-profile.cpu-cycles.__fget_light.do_sys_poll.sys_poll.system_call_fastpath
0.00 Â 0% +Inf% 33.68 Â 8% perf-profile.cpu-cycles.fput.do_sys_poll.sys_poll.system_call_fastpath
25 Â 37% -33.0% 16 Â 4% sched_debug.cfs_rq[15]:/.runnable_load_avg
37159 Â 9% +20.7% 44850 Â 16% sched_debug.cfs_rq[24]:/.exec_clock
830542 Â 10% +18.8% 986752 Â 15% sched_debug.cfs_rq[24]:/.min_vruntime
18120 Â 8% -17.8% 14897 Â 19% sched_debug.cfs_rq[8]:/.avg->runnable_avg_sum
394 Â 8% -17.8% 323 Â 19% sched_debug.cfs_rq[8]:/.tg_runnable_contrib
26 Â 43% -36.2% 16 Â 4% sched_debug.cpu#15.cpu_load[1]
23 Â 30% -28.3% 16 Â 5% sched_debug.cpu#15.cpu_load[2]
1564 Â 21% +73.2% 2709 Â 47% sched_debug.cpu#20.nr_switches
617 Â 27% +99.6% 1232 Â 48% sched_debug.cpu#20.ttwu_count
2164 Â 20% -28.2% 1553 Â 35% sched_debug.cpu#25.ttwu_count
1902 Â 21% -26.6% 1397 Â 36% sched_debug.cpu#25.sched_goidle
1947 Â 35% -62.6% 728 Â 35% sched_debug.cpu#26.ttwu_local
5620 Â 23% -41.6% 3282 Â 15% sched_debug.cpu#26.nr_switches
2595 Â 26% -43.5% 1465 Â 19% sched_debug.cpu#26.ttwu_count
653 Â 9% +76.3% 1151 Â 34% sched_debug.cpu#31.ttwu_local
1162 Â 10% +34.5% 1563 Â 22% sched_debug.cpu#31.sched_goidle
2885 Â 9% +34.5% 3880 Â 22% sched_debug.cpu#31.nr_switches

lkp-sb03: Sandy Bridge-EP
Memory: 64G




perf-profile.cpu-cycles.do_sys_poll.sys_poll.system_call_fastpath

95 ++---------------------------------------------------------------------+
| O O O O O O O O O O O O O O O O O O O O |
90 ++ O O O |
85 ++ |
O |
80 ++ O
| |
75 ++ |
| |
70 ++ |
65 ++ |
| |
60 *+.*..*..*..*.. .*..*..*..*..*..*..*.*..*..*..*..*..*..*..* |
| *. |
55 ++---------------------------------------------------------------------+


perf-profile.cpu-cycles.fput.do_sys_poll.sys_poll.system_call_fastpath

37 ++---------------------------------------------------------------------+
| O |
36 ++ O O |
35 ++ O O O O O O O O O O O O O O |
| O O O |
34 ++ O O |
33 ++ O |
| |
32 ++ |
31 ++ |
| |
30 O+ |
29 ++ |
| O
28 ++---------------------------------------------------------------------+


perf-profile.cpu-cycles.fput.sys_poll.system_call_fastpath

35 ++---------------------------------------------------------------------+
*..*..*..*..*.. .*..*..*..*..*..*..*.*..*..*..*..*..*..*..* |
30 ++ *. |
| |
25 ++ |
| |
20 ++ |
| |
15 ++ |
| |
10 ++ |
| |
5 ++ |
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
0 ++---------------------------------------------------------------------+


will-it-scale.per_thread_ops

290000 ++-----------------------------------------------------------------+
| |
285000 ++ *..*.. |
| : |
*..*..*.*..*.. .*..* : * *..*..*..*.*.. .*..* |
280000 ++ *. + : + : *. |
| * + : |
275000 ++ * |
| |
270000 ++ |
| |
| O O O O O O O O |
265000 ++ O O O O O O O O O O O O |
O O O O
260000 ++---------------O-------------------------------------------------+


perf-profile.cpu-cycles.__fget_light.do_sys_poll.sys_poll.system_call_fastpath

3 ++--------------------------------------------------------------------+
*.. .*..*..*..*.. .*..*..*..*.. .*..*..*.. .*..*..*.*..* |
2.5 ++ *. * *. *. |
| |
| |
2 ++ |
| |
1.5 ++ |
| |
1 ++ |
| |
| |
0.5 ++ |
| |
0 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


perf-profile.cpu-cycles.__fdget.sys_poll.system_call_fastpath

3 ++--------------------------------------------------------------------+
*..*..*..*..*..*..*.*..*..*..*..*..*.. .*.. .* |
2.5 ++ *..*..*..*. *.*. |
| |
| |
2 ++ |
| |
1.5 ++ |
| |
1 ++ |
| |
| |
0.5 ++ |
| |
0 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

[*] 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: will-it-scale
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:
commit: 2ed2a023ae54e95a081668b8c3dbb97bebfe3aa5
model: Sandy Bridge-EP
memory: 64G
hdd_partitions: "/dev/disk/by-id/ata-WDC_WD10EARS-00Y5B1_WD-WCAV5F059499-part3"
swap_partitions:
rootfs_partition: "/dev/disk/by-id/ata-WDC_WD10EARS-00Y5B1_WD-WCAV5F059499-part4"
perf-profile:
freq: 800
will-it-scale:
test: poll2
testbox: lkp-sb03
tbox_group: lkp-sb03
kconfig: x86_64-rhel
enqueue_time: 2015-03-09 00:35:59.009246505 +08:00
head_commit: 2ed2a023ae54e95a081668b8c3dbb97bebfe3aa5
base_commit: 13a7a6ac0a11197edcd0f756a035f472b42cdf8b
branch: linux-devel/devel-hourly-2015030905
kernel: "/kernel/x86_64-rhel/2ed2a023ae54e95a081668b8c3dbb97bebfe3aa5/vmlinuz-4.0.0-rc2-00977-g2ed2a02"
user: lkp
queue: cyclic
rootfs: debian-x86_64-2015-02-07.cgz
result_root: "/result/lkp-sb03/will-it-scale/poll2/debian-x86_64-2015-02-07.cgz/x86_64-rhel/2ed2a023ae54e95a081668b8c3dbb97bebfe3aa5/0"
job_file: "/lkp/scheduled/lkp-sb03/cyclic_will-it-scale-poll2-x86_64-rhel-HEAD-2ed2a023ae54e95a081668b8c3dbb97bebfe3aa5-0-20150309-58098-1cjbbyh.yaml"
dequeue_time: 2015-03-09 08:13:02.769544699 +08:00
nr_cpu: "$(nproc)"
job_state: finished
loadavg: 21.58 11.95 4.85 1/309 9795
start_time: '1425860035'
end_time: '1425860345'
version: "/lkp/lkp/.src-20150308-175746"
./runtest.py poll2 25 both 1 8 16 24 32
_______________________________________________
LKP mailing list
LKP@xxxxxxxxxxxxxxx