Hi,
We noticed the below changes on(NOTE: I'm not sure the bisect is correct
or not, here I report it out JFYI).
git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
commit 5fcb864ef90df093d964171539c87ffa0ab49f0f ("x86, irq, ACPI: Implement interfaces to support ACPI based IOAPIC hot-removal")
testbox/testcase/testparams: lkp-nex06/vm-scalability/performance-300s-small-allocs-mt
ff6213974cd90e1e 5fcb864ef90df093d964171539
---------------- --------------------------
fail:runs %reproduction fail:runs
| | |
:5 20% 1:5 kmsg.CE:hpet_increased_min_delta_ns_to#nsec
%stddev %change %stddev
\ | \
315326 ? 0% -3.3% 304841 ? 0% vm-scalability.throughput
11.82 ? 0% +12.9% 13.34 ? 0% turbostat.%c0
1.34 ? 0% +9.4% 1.46 ? 0% turbostat.GHz
12 ? 47% +78.7% 21 ? 32% sched_debug.cfs_rq[29]:/.load
113 ? 26% +86.3% 212 ? 28% sched_debug.cfs_rq[39]:/.tg_load_contrib
106 ? 28% +89.5% 202 ? 30% sched_debug.cfs_rq[39]:/.blocked_load_avg
66 ? 23% +120.6% 145 ? 29% sched_debug.cfs_rq[40]:/.blocked_load_avg
70 ? 23% +113.0% 150 ? 29% sched_debug.cfs_rq[40]:/.tg_load_contrib
10145 ? 23% -38.3% 6255 ? 35% numa-meminfo.node1.AnonPages
2535 ? 23% -38.3% 1564 ? 35% numa-vmstat.node1.nr_anon_pages
605 ? 16% -22.0% 471 ? 5% sched_debug.cpu#58.nr_uninterruptible
58904 ? 7% -13.8% 50762 ? 7% sched_debug.cfs_rq[0]:/.min_vruntime
481299 ? 8% -13.4% 416975 ? 7% sched_debug.cpu#0.sched_count
409009 ? 11% -15.7% 344638 ? 2% sched_debug.cpu#4.sched_count
52022 ? 10% -16.1% 43623 ? 2% sched_debug.cfs_rq[4]:/.min_vruntime
68 ? 3% -12.2% 60 ? 3% sched_debug.cfs_rq[4]:/.tg_runnable_contrib
3175 ? 3% -12.1% 2791 ? 3% sched_debug.cfs_rq[4]:/.avg->runnable_avg_sum
50060 ? 6% -12.3% 43914 ? 4% sched_debug.cfs_rq[29]:/.min_vruntime
1751 ? 12% -15.5% 1480 ? 6% sched_debug.cpu#63.nr_uninterruptible
2967 ? 6% -13.7% 2562 ? 4% sched_debug.cfs_rq[37]:/.avg->runnable_avg_sum
63 ? 6% -13.8% 55 ? 4% sched_debug.cfs_rq[37]:/.tg_runnable_contrib
1.07 ? 2% -10.9% 0.95 ? 3% perf-profile.cpu-cycles.tick_nohz_restart.tick_nohz_idle_exit.cpu_startup_entry.start_secondary
1.64 ? 2% -8.4% 1.50 ? 4% perf-profile.cpu-cycles.__tick_nohz_idle_enter.tick_nohz_idle_enter.cpu_startup_entry.start_secondary
35173 ? 5% -9.1% 31983 ? 3% sched_debug.cfs_rq[56]:/.min_vruntime
1.41 ? 2% -8.3% 1.29 ? 4% perf-profile.cpu-cycles.tick_nohz_stop_sched_tick.__tick_nohz_idle_enter.tick_nohz_idle_enter.cpu_startup_entry.start_secondary
1.63 ? 1% -9.3% 1.48 ? 3% perf-profile.cpu-cycles.tick_nohz_idle_exit.cpu_startup_entry.start_secondary
45161 ? 11% -12.8% 39358 ? 4% sched_debug.cfs_rq[25]:/.min_vruntime
39201 ? 5% +17.3% 45969 ? 18% sched_debug.cfs_rq[8]:/.min_vruntime
21071502 ? 0% -3.3% 20379730 ? 0% time.minor_page_faults
299 ? 0% -3.1% 290 ? 0% time.user_time
21763267 ? 0% -3.3% 21055329 ? 0% time.voluntary_context_switches
142199 ? 0% -3.1% 137732 ? 0% vmstat.system.cs
737 ? 0% -2.1% 721 ? 1% time.system_time
341 ? 0% -2.5% 333 ? 0% time.percent_of_cpu_this_job_got
lkp-nex06: Nehalem-EX
Memory: 64G
turbostat.%c0
14 ++-------------------------------------------------------------------+
| O |
| O O O O O |
13.5 O+ O O O O O O O |
| O O O O O |
| O O O
13 ++ |
| |
12.5 ++ |
| |
| |
12 *+.*...*.. .*... |
| *..*. *.. .*...*..*..*..*... .*..* |
| *. *. |
11.5 ++-------------------------------------------------------------------+
turbostat.GHz
1.48 ++-------------------------------------------------------------------+
O O O O O O O O O O O O |
1.46 ++ O O O O O O O O O O
1.44 ++ |
| |
1.42 ++ |
| |
1.4 ++ |
| |
1.38 ++ |
1.36 ++ |
| |
1.34 ++ .*...*.. .*..*... .*.. |
*..*...*..*..*. *..*...*..*. *. * |
1.32 ++-------------------------------------------------------------------+
[*] 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.
--yliu
On Mon, 24 Nov 2014, Yuanhan Liu wrote:
>
> We noticed the below changes on(NOTE: I'm not sure the bisect is correct
> or not, here I report it out JFYI).
>
> git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
> commit 5fcb864ef90df093d964171539c87ffa0ab49f0f ("x86, irq, ACPI: Implement interfaces to support ACPI based IOAPIC hot-removal")
>
I really doubt that this commit has anything to do with the observed
behaviour change.
The code in question is only executed at boot time or when a IOAPIC is
physically hotplugged.
Thanks,
tglx