Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1360356 > unrolled thread
| Started by | "Huang\, Ying" <ying.huang@intel.com> |
|---|---|
| First post | 2016-03-18 07:20 +0100 |
| Last post | 2016-03-29 09:50 +0200 |
| Articles | 10 — 4 participants |
Back to article view | Back to linux.kernel
This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by
below is the oldest one visible, not the original post.
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops "Huang\, Ying" <ying.huang@intel.com> - 2016-03-18 07:20 +0100
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops Thomas Gleixner <tglx@linutronix.de> - 2016-03-18 09:40 +0100
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops "Huang\, Ying" <ying.huang@intel.com> - 2016-03-21 04:00 +0100
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops Thomas Gleixner <tglx@linutronix.de> - 2016-03-21 09:10 +0100
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops "Huang\, Ying" <ying.huang@intel.com> - 2016-03-21 09:50 +0100
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops Darren Hart <dvhart@infradead.org> - 2016-03-28 19:20 +0200
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops "Huang\, Ying" <ying.huang@intel.com> - 2016-03-29 03:20 +0200
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops Darren Hart <dvhart@infradead.org> - 2016-03-29 07:20 +0200
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops "Huang\, Ying" <ying.huang@intel.com> - 2016-03-29 08:00 +0200
Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops Peter Zijlstra <peterz@infradead.org> - 2016-03-29 09:50 +0200
| From | "Huang\, Ying" <ying.huang@intel.com> |
|---|---|
| Date | 2016-03-18 07:20 +0100 |
| Subject | Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops |
| Message-ID | <rdVVo-1Lv-9@gated-at.bofh.it> |
Sorry, for late.
Ingo Molnar <mingo@kernel.org> writes:
> * kernel test robot <ying.huang@linux.intel.com> wrote:
>
>> FYI, we noticed the below changes on
>>
>> https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
>> commit 65d8fc777f6dcfee12785c057a6b57f679641c90 ("futex: Remove requirement for
>> lock_page() in get_futex_key()")
>
> I have asked for this before, but let me try again: could you _PLEASE_ make these
> emails more readable?
>
> For example what are the 'below changes'? Changes in the profile output? Profiles
> always change from run to run, so that alone is not informative.
>
> Also, there are a lot of changes - which ones prompted the email to be generated?
Usually we will put most important change we think in the subject of the
mail, for this email, it is,
+25.6% will-it-scale.per_process_ops
and, we try to put most important changes at the top of the comparison
result below. That is the will-it-scale.xxx below.
We are thinking about how to improve this. You input is valuable for
us. We are thinking change the "below changes" line to something like
below.
FYI, we noticed the +25.6% will-it-scale.per_process_ops improvement on
...
Does this looks better?
> All in one, this email is hard to parse, because it just dumps a lot of
> information with very little explanatory structure for someone not versed in their
> format. Please try to create an easy to parse 'story' that leads the reader
> towards what you want these emails to tell - not just a raw dump of seemingly
> unconnected pieces of data ...
Which kind of story? Something like, we tested some benchmark on some
machine, which triggered some regression, we do bisect, and find the
first bad commit is xxx. In addition to benchmark result, we collected
some other information, hope they are helpful for you.
We just try to help. Sorry for confusion. We try to provide all
information needed to describe the change and help for root causing the
changes. But we know, we are still far from there, so your input is
important for us to improve our test report. Which part of the report
do you think should be changed firstly? The overall structure? Or the
data format?
Best Regards,
Huang, Ying
> Thanks,
>
> Ingo
>
>>
>>
>> =========================================================================================
>> compiler/cpufreq_governor/kconfig/rootfs/tbox_group/test/testcase:
>> gcc-4.9/performance/x86_64-rhel/debian-x86_64-2015-02-07.cgz/lkp-sbx04/futex1/will-it-scale
>>
>> commit:
>> 8ad7b378d0d016309014cae0f640434bca7b5e11
>> 65d8fc777f6dcfee12785c057a6b57f679641c90
>>
>> 8ad7b378d0d01630 65d8fc777f6dcfee12785c057a
>> ---------------- --------------------------
>> %stddev %change %stddev
>> \ | \
>> 5076304 . 0% +25.6% 6374220 . 0% will-it-scale.per_process_ops
>> 1194117 . 0% +14.4% 1366153 . 1% will-it-scale.per_thread_ops
>> 0.58 . 0% -2.0% 0.57 . 0% will-it-scale.scalability
>> 6820 . 0% -19.6% 5483 . 15% meminfo.AnonHugePages
>> 2652 . 5% -10.4% 2375 . 2% vmstat.system.cs
>> 2848 . 32% +141.2% 6870 . 65% numa-meminfo.node1.Active(anon)
>> 2832 . 31% +57.6% 4465 . 27% numa-meminfo.node1.AnonPages
>> 15018 . 12% -23.3% 11515 . 15% numa-meminfo.node2.AnonPages
>> 1214 . 14% -22.8% 936.75 . 20% numa-meminfo.node3.PageTables
>> 712.25 . 32% +141.2% 1718 . 65% numa-vmstat.node1.nr_active_anon
>> 708.25 . 31% +57.7% 1116 . 27% numa-vmstat.node1.nr_anon_pages
>> 3754 . 12% -23.3% 2879 . 15% numa-vmstat.node2.nr_anon_pages
>> 304.75 . 14% -23.1% 234.50 . 20% numa-vmstat.node3.nr_page_table_pages
>> 3.53 . 1% -100.0% 0.00 . -1%
> perf-profile.cycles.___might_sleep.__might_sleep.get_futex_key.futex_wake.do_futex
>> 4.34 . 1% -100.0% 0.00 . -1%
> perf-profile.cycles.__might_sleep.get_futex_key.futex_wake.do_futex.sys_futex
>> 1.27 . 3% -100.0% 0.00 . -1%
> perf-profile.cycles.__wake_up_bit.unlock_page.get_futex_key.futex_wake.do_futex
>> 4.36 . 1% +29.6% 5.65 . 1%
> perf-profile.cycles.drop_futex_key_refs.isra.12.futex_wake.do_futex.sys_futex.entry_SYSCALL_64_fastpath
>> 6.69 . 1% +28.1% 8.57 . 0% perf-profile.cycles.entry_SYSCALL_64
>> 6.73 . 0% +30.6% 8.79 . 0% perf-profile.cycles.entry_SYSCALL_64_after_swapgs
>> 74.21 . 0% -11.0% 66.06 . 0%
> perf-profile.cycles.futex_wake.do_futex.sys_futex.entry_SYSCALL_64_fastpath
>> 59.05 . 0% -21.4% 46.40 . 0%
> perf-profile.cycles.get_futex_key.futex_wake.do_futex.sys_futex.entry_SYSCALL_64_fastpath
>> 4.12 . 0% +78.5% 7.36 . 1%
> perf-profile.cycles.get_futex_key_refs.isra.11.futex_wake.do_futex.sys_futex.entry_SYSCALL_64_fastpath
>> 2.27 . 3% +24.1% 2.82 . 4%
> perf-profile.cycles.get_user_pages_fast.futex_wake.do_futex.sys_futex.entry_SYSCALL_64_fastpath
>> 26.95 . 0% +30.0% 35.04 . 1%
> perf-profile.cycles.get_user_pages_fast.get_futex_key.futex_wake.do_futex.sys_futex
>> 13.43 . 0% +27.2% 17.09 . 1%
> perf-profile.cycles.gup_pte_range.gup_pud_range.get_user_pages_fast.get_futex_key.futex_wake
>> 19.66 . 1% +28.4% 25.24 . 0%
> perf-profile.cycles.gup_pud_range.get_user_pages_fast.get_futex_key.futex_wake.do_futex
>> 4.33 . 1% +37.0% 5.93 . 4%
> perf-profile.cycles.hash_futex.do_futex.sys_futex.entry_SYSCALL_64_fastpath
>> 13.59 . 0% -100.0% 0.00 . -1%
> perf-profile.cycles.unlock_page.get_futex_key.futex_wake.do_futex.sys_futex
>> 15160 . 19% -34.8% 9883 . 0% sched_debug.cfs_rq:/.exec_clock.min
>> 27.25 . 15% -37.6% 17.00 . 8% sched_debug.cfs_rq:/.load_avg.7
>> 21.00 . 38% -27.4% 15.25 . 2% sched_debug.cpu.cpu_load[2].1
>> 21.00 . 38% -27.4% 15.25 . 2% sched_debug.cpu.cpu_load[3].1
>> 21.00 . 38% -27.4% 15.25 . 2% sched_debug.cpu.cpu_load[4].1
>> 1790 . 0% +42.4% 2549 . 45% sched_debug.cpu.curr->pid.21
>> 50033 . 4% -6.8% 46622 . 4% sched_debug.cpu.nr_load_updates.29
>> 4398 . 42% +103.5% 8949 . 23% sched_debug.cpu.nr_switches.11
>> 7452 . 34% +111.3% 15744 . 54% sched_debug.cpu.nr_switches.20
>> 3739 . 13% +213.5% 11723 . 40% sched_debug.cpu.nr_switches.23
>> 1648 . 53% +96.5% 3239 . 63% sched_debug.cpu.nr_switches.51
>> 0.25 .519% -1300.0% -3.00 .-52% sched_debug.cpu.nr_uninterruptible.24
>> 8632 . 16% -32.5% 5823 . 19% sched_debug.cpu.sched_count.1
>> 5091 . 36% +137.5% 12092 . 31% sched_debug.cpu.sched_count.11
>> 12453 . 90% -74.6% 3159 . 24% sched_debug.cpu.sched_count.2
>> 7782 . 32% +118.2% 16977 . 46% sched_debug.cpu.sched_count.20
>> 2665 . 48% -49.8% 1337 . 30% sched_debug.cpu.sched_count.32
>> 1365 . 11% -14.0% 1174 . 3% sched_debug.cpu.sched_count.45
>> 1693 . 51% +147.7% 4193 . 42% sched_debug.cpu.sched_count.51
>> 5023 . 57% -51.5% 2434 . 43% sched_debug.cpu.sched_count.57
>> 1705 . 16% +129.6% 3915 . 48% sched_debug.cpu.sched_goidle.23
>> 536.25 . 14% -18.7% 435.75 . 2% sched_debug.cpu.sched_goidle.45
>> 1228 . 19% -27.3% 892.50 . 17% sched_debug.cpu.sched_goidle.5
>> 1919 . 55% +88.5% 3617 . 37% sched_debug.cpu.ttwu_count.11
>> 7699 . 35% -43.7% 4335 . 43% sched_debug.cpu.ttwu_count.24
>> 5380 . 36% -45.6% 2926 . 18% sched_debug.cpu.ttwu_count.30
>> 563.25 . 20% +140.3% 1353 . 38% sched_debug.cpu.ttwu_local.11
>> 4297 . 46% -49.1% 2186 . 39% sched_debug.cpu.ttwu_local.24
>> 2828 . 47% -47.8% 1475 . 34% sched_debug.cpu.ttwu_local.27
>> 3243 . 36% -54.3% 1482 . 32% sched_debug.cpu.ttwu_local.30
>> 199.25 . 6% +100.6% 399.75 . 32% sched_debug.cpu.ttwu_local.44
>> 1158 . 64% -67.3% 379.00 . 46% sched_debug.cpu.ttwu_local.54
>> 242.25 . 21% +51.0% 365.75 . 19% sched_debug.cpu.ttwu_local.55
>> 1009 . 26% -50.8% 496.50 . 44% sched_debug.cpu.ttwu_local.59
>> 1736 . 53% -67.8% 559.25 . 22% sched_debug.cpu.ttwu_local.9
>>
>>
>> lkp-sbx04: Sandy Bridge-EX
>> Memory: 64G
>>
>>
>> perf-profile.cycles.futex_wake.do_futex.sys_futex.entry_SYSCALL_64_fastpath
>>
>> 76 ++---------------------------------------------------------------------+
>> | |
>> 74 ++ .*.. .*..*..*.. .*.. .*.. .*.. .*.. .*..*..*..* |
>> *..*. *..*..*. *..*. *.*. *. *. *. |
>> | |
>> 72 ++ |
>> | |
>> 70 ++ |
>> | |
>> 68 ++ |
>> | |
>> | O O O O O O O O O O O |
>> 66 O+ O O O O O
>> | O O O O O O O |
>> 64 ++-------------O-------------------------------------------------------+
>>
>>
>>
>> will-it-scale.per_process_ops
>>
>> 6.6e+06 O+----O-O--O------------------------------------------------------+
>> | O O O O O |
>> 6.4e+06 ++ O O O O O O O O O O O O O O O O
>> | |
>> 6.2e+06 ++ |
>> 6e+06 ++ |
>> | |
>> 5.8e+06 ++ |
>> | |
>> 5.6e+06 ++ |
>> 5.4e+06 ++ |
>> | |
>> 5.2e+06 ++ |
>> *..*..*.*..*..*..*.*..*..*..*.*..*..*..*.*..*..*..*.*..*..*..*.* |
>> 5e+06 ++----------------------------------------------------------------+
>>
>>
>> [*] bisect-good sample
>> [O] bisect-bad sample
>>
>> 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
>>
>>
>> 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
>
>> ---
>> LKP_SERVER: inn
>> LKP_CGI_PORT: 80
>> LKP_CIFS_PORT: 139
>> testcase: will-it-scale
>> default-monitors:
>> wait: activate-monitor
>> kmsg:
>> uptime:
>> iostat:
>> heartbeat:
>> vmstat:
>> numa-numastat:
>> numa-vmstat:
>> numa-meminfo:
>> proc-vmstat:
>> proc-stat:
>> interval: 10
>> meminfo:
>> slabinfo:
>> interrupts:
>> lock_stat:
>> latency_stats:
>> softirqs:
>> bdi_dev_mapping:
>> diskstats:
>> nfsstat:
>> cpuidle:
>> cpufreq-stats:
>> turbostat:
>> pmeter:
>> sched_debug:
>> interval: 60
>> cpufreq_governor: performance
>> default-watchdogs:
>> oom-killer:
>> watchdog:
>> commit: 65d8fc777f6dcfee12785c057a6b57f679641c90
>> model: Sandy Bridge-EX
>> nr_cpu: 64
>> memory: 64G
>> nr_ssd_partitions: 7
>> ssd_partitions: "/dev/disk/by-id/ata-INTEL_SSDSC2*-part1"
>> swap_partitions:
>> category: benchmark
>> perf-profile:
>> freq: 800
>> will-it-scale:
>> test: futex1
>> queue: bisect
>> testbox: lkp-sbx04
>> tbox_group: lkp-sbx04
>> kconfig: x86_64-rhel
>> enqueue_time: 2016-02-28 23:45:52.199165563 +08:00
>> compiler: gcc-4.9
>> rootfs: debian-x86_64-2015-02-07.cgz
>> id: 6b2c2bd744dd898009648cb82de7e0ba77de33f1
>> user: lkp
>> head_commit: ed520c327c4259ec08b1677023087f658329b961
>> base_commit: 81f70ba233d5f660e1ea5fe23260ee323af5d53a
>> branch: linux-devel/devel-hourly-2016022811
>> result_root:
> "/result/will-it-scale/performance-futex1/lkp-sbx04/debian-x86_64-2015-02-07.cgz/x86_64-rhel/gcc-4.9/65d8fc777f6dcfee12785c057a6b57f679641c90/0"
>> job_file:
> "/lkp/scheduled/lkp-sbx04/bisect_will-it-scale-performance-futex1-debian-x86_64-2015-02-07.cgz-x86_64-rhel-65d8fc777f6dcfee12785c057a6b57f679641c90-20160228-23650-17c4qc1-0.yaml"
>> max_uptime: 1500
>> initrd: "/osimage/debian/debian-x86_64-2015-02-07.cgz"
>> bootloader_append:
>> - root=/dev/ram0
>> - user=lkp
>> -
> job=/lkp/scheduled/lkp-sbx04/bisect_will-it-scale-performance-futex1-debian-x86_64-2015-02-07.cgz-x86_64-rhel-65d8fc777f6dcfee12785c057a6b57f679641c90-20160228-23650-17c4qc1-0.yaml
>> - ARCH=x86_64
>> - kconfig=x86_64-rhel
>> - branch=linux-devel/devel-hourly-2016022811
>> - commit=65d8fc777f6dcfee12785c057a6b57f679641c90
>> -
> BOOT_IMAGE=/pkg/linux/x86_64-rhel/gcc-4.9/65d8fc777f6dcfee12785c057a6b57f679641c90/vmlinuz-4.5.0-rc3-00235-g65d8fc7
>> - max_uptime=1500
>> -
> RESULT_ROOT=/result/will-it-scale/performance-futex1/lkp-sbx04/debian-x86_64-2015-02-07.cgz/x86_64-rhel/gcc-4.9/65d8fc777f6dcfee12785c057a6b57f679641c90/0
>> - LKP_SERVER=inn
>> - |2-
>>
>>
>> earlyprintk=ttyS0,115200 systemd.log_level=err
>> debug apic=debug sysrq_always_enabled rcupdate.rcu_cpu_stall_timeout=100
>> panic=-1 softlockup_panic=1 nmi_watchdog=panic oops=panic load_ramdisk=2 prompt_ramdisk=0
>> console=ttyS0,115200 console=tty0 vga=normal
>>
>> rw
>> lkp_initrd: "/lkp/lkp/lkp-x86_64.cgz"
>> modules_initrd: "/pkg/linux/x86_64-rhel/gcc-4.9/65d8fc777f6dcfee12785c057a6b57f679641c90/modules.cgz"
>> bm_initrd:
> "/osimage/deps/debian-x86_64-2015-02-07.cgz/lkp.cgz,/osimage/deps/debian-x86_64-2015-02-07.cgz/run-ipconfig.cgz,/osimage/deps/debian-x86_64-2015-02-07.cgz/turbostat.cgz,/lkp/benchmarks/turbostat.cgz,/osimage/deps/debian-x86_64-2015-02-07.cgz/will-it-scale.cgz,/lkp/benchmarks/will-it-scale.cgz,/lkp/benchmarks/will-it-scale-x86_64.cgz"
>> linux_headers_initrd: "/pkg/linux/x86_64-rhel/gcc-4.9/65d8fc777f6dcfee12785c057a6b57f679641c90/linux-headers.cgz"
>> repeat_to: 2
>> kernel: "/pkg/linux/x86_64-rhel/gcc-4.9/65d8fc777f6dcfee12785c057a6b57f679641c90/vmlinuz-4.5.0-rc3-00235-g65d8fc7"
>> dequeue_time: 2016-02-28 23:46:33.938915178 +08:00
>> job_state: finished
>> loadavg: 45.27 20.12 7.84 2/649 11559
>> start_time: '1456674445'
>> end_time: '1456674754'
>> version: "/lkp/lkp/.src-20160226-194908"
>
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu10/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu11/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu12/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu13/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu14/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu15/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu16/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu17/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu18/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu19/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu2/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu20/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu21/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu22/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu23/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu24/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu25/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu26/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu27/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu28/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu29/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu3/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu30/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu31/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu32/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu33/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu34/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu35/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu36/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu37/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu38/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu39/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu4/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu40/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu41/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu42/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu43/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu44/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu45/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu46/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu47/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu48/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu49/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu5/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu50/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu51/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu52/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu53/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu54/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu55/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu56/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu57/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu58/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu59/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu6/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu60/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu61/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu62/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu63/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu7/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu8/cpufreq/scaling_governor
>> 2016-02-28 23:47:24 echo performance > /sys/devices/system/cpu/cpu9/cpufreq/scaling_governor
>> 2016-02-28 23:47:25 ./runtest.py futex1 16 both 1 8 16 24 32 48 64
>
> _______________________________________________
> LKP mailing list
> LKP@lists.01.org
> https://lists.01.org/mailman/listinfo/lkp
[toc] | [next] | [standalone]
| From | Thomas Gleixner <tglx@linutronix.de> |
|---|---|
| Date | 2016-03-18 09:40 +0100 |
| Subject | Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops |
| Message-ID | <rdY6S-36M-13@gated-at.bofh.it> |
| In reply to | #1360356 |
On Fri, 18 Mar 2016, Huang, Ying wrote:
> Usually we will put most important change we think in the subject of the
> mail, for this email, it is,
>
> +25.6% will-it-scale.per_process_ops
That is confusing on it's own, because the reader does not know at all whether
this is an improvement or a regression.
So something like this might be useful:
Subject: subsystem: 12digitsha1: 25% performance improvement
or in some other case
Subject: subsystem: 12digitsha1: 25% performance regression
So in the latter case I will look into that mail immediately. The improvement
one can wait until I have cared about urgent stuff.
In the subject line it is pretty much irrelevant which foo-bla-ops test has
produced that result. It really does not matter. If it's a regression, it's
urgent. If it's an improvement it's informal and it can wait to be read.
So in that case it would be:
futex: 65d8fc777f6d: 25% performance improvement
You can grab the subsystem prefix from the commit.
> and, we try to put most important changes at the top of the comparison
> result below. That is the will-it-scale.xxx below.
>
> We are thinking about how to improve this. You input is valuable for
> us. We are thinking change the "below changes" line to something like
> below.
>
> FYI, we noticed the +25.6% will-it-scale.per_process_ops improvement on
> ...
>
> Does this looks better?
A bit, but it still does not tell me much. It's completely non obvious what
'will-it-scale.per_process_ops' means. Let me give you an example how a useful
and easy to understand summary of the change could look like:
FYI, we noticed 25.6% performance improvement due to commit
65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()"
in the will-it-scale.per_process_ops test.
will-it-scale.per_process_ops tests the futex operations for process shared
futexes (Or whatever that test really does).
The commit has no significant impact on any other test in the test suite.
So those few lines tell precisely what this is about. It's something I already
expected, so I really can skip the rest of the mail unless I'm interested in
reproducing the result.
Now lets look at a performance regression.
Subject: futex: 65d8fc777f6d: 25% performance regression
FYI, we noticed a 25.2% performance regression due to commit
65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()"
in the will-it-scale.per_process_ops test.
will-it-scale.per_process_ops tests the futex operations for process shared
futexes (Or whatever that test really does).
The commit has no significant impact on any other test in the test suite.
In that case I will certainly be interested how to reproduce that test. So I
need the following information:
Machine description: Intel IvyBridge 2 sockets, 32 cores, 64G RAM
Config file: http://wherever.you.store/your/results/test-nr/config
Test: http://wherever.you.store/your/tests/will-it-scale.per_process_ops.tar.bz2
That tarball should contain:
README
test_script.sh
test_binary
README should tell:
will-it-scale.per_process_ops
Short explanation of the test
Preliminaries:
- perf
- whatever
So that allows me to reproduce that test more or less with no effort. And
that's the really important part.
You can provide nice charts and full comparison tables for all tests on a web
site for those who are interested in large stats and pretty charts.
Full results: http://wherever.you.store/your/results/test-nr/results
So now lets look at a scenario where that commit results in a performance
improvement on will-it-scale.per_process_ops and a regression on
will-it-scale.per_task.
Subject: futex: 65d8fc777f6d: 25% performance regression
FYI, we noticed a 25.2% performance regression due to commit
65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()"
in the will-it-scale.per_task test.
will-it-scale.per_tasks tests the futex operations for process private
futexes (Or whatever that test really does).
The commit has significant impact on the following tests in the test suite:
will-it-scale.per_process_ops: 25% improvement
The 25% improvement is really not interesting. What's interesting is the
regression. That's what people need to look at. You still can provide the
information about all the tests including the 25% improvement data on
Full results: http://wherever.you.store/your/results/test-nr/results
Now, if you have commits which have mixed impact on various tests, then it's
important to point out the most relevant issue clearly in the
summary. I.e. you need to filter the tests for the maximum
regression/improvement and use that as the main information. You still can
provide the information about the impact on other tests in a very condensed
form.
The commit has significant impact on the following tests in the test suite:
test1: 0.5% Regression
test2: 2.0% Improvement
....
testN: 0.2% Regression
That's useful information, but it might be completely irrelevant if the
primary impact is, e.g. a 10% Regression. Then it's nice to know, that it gave
a 2% improvement on test2, but that's about it. The important information is
the 10% regression, which needs to be addressed urgently.
Hope that helps.
Thanks,
tglx
[toc] | [prev] | [next] | [standalone]
| From | "Huang\, Ying" <ying.huang@intel.com> |
|---|---|
| Date | 2016-03-21 04:00 +0100 |
| Message-ID | <reYeu-7GT-5@gated-at.bofh.it> |
| In reply to | #1360423 |
Hi, Thomas, Thanks a lot for your valuable input! Thomas Gleixner <tglx@linutronix.de> writes: > On Fri, 18 Mar 2016, Huang, Ying wrote: >> Usually we will put most important change we think in the subject of the >> mail, for this email, it is, >> >> +25.6% will-it-scale.per_process_ops > > That is confusing on it's own, because the reader does not know at all whether > this is an improvement or a regression. > > So something like this might be useful: > > Subject: subsystem: 12digitsha1: 25% performance improvement > > or in some other case > > Subject: subsystem: 12digitsha1: 25% performance regression > > So in the latter case I will look into that mail immediately. The improvement > one can wait until I have cared about urgent stuff. > > In the subject line it is pretty much irrelevant which foo-bla-ops test has > produced that result. It really does not matter. If it's a regression, it's > urgent. If it's an improvement it's informal and it can wait to be read. > > So in that case it would be: > > futex: 65d8fc777f6d: 25% performance improvement > > You can grab the subsystem prefix from the commit. We will include regression/improvement information in subject at least. >> and, we try to put most important changes at the top of the comparison >> result below. That is the will-it-scale.xxx below. >> >> We are thinking about how to improve this. You input is valuable for >> us. We are thinking change the "below changes" line to something like >> below. >> >> FYI, we noticed the +25.6% will-it-scale.per_process_ops improvement on >> ... >> >> Does this looks better? > > A bit, but it still does not tell me much. It's completely non obvious what > 'will-it-scale.per_process_ops' means. will-it-scale is a test suite, per_process_ops is one of its results. That is the convention used in original report. > Let me give you an example how a useful > and easy to understand summary of the change could look like: > > > FYI, we noticed 25.6% performance improvement due to commit > > 65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()" > > in the will-it-scale.per_process_ops test. > > will-it-scale.per_process_ops tests the futex operations for process shared > futexes (Or whatever that test really does). There is a futex sub test case for will-it-scale test suite. But I got your point, we need some description for the test case. If email is not too limited for the full description, we will put it in some web site and include short description and link to the full description in email. > The commit has no significant impact on any other test in the test suite. Sorry, we have no enough machine power to test all test cases for each bisect result. So we will have no such information until we find a way to do that. > So those few lines tell precisely what this is about. It's something I already > expected, so I really can skip the rest of the mail unless I'm interested in > reproducing the result. We will put important information at first of the email. And details later. Better to have clear mark. So people can get important information and ignore the details if they don't want. > Now lets look at a performance regression. > > Subject: futex: 65d8fc777f6d: 25% performance regression > > FYI, we noticed a 25.2% performance regression due to commit > > 65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()" > > in the will-it-scale.per_process_ops test. > > will-it-scale.per_process_ops tests the futex operations for process shared > futexes (Or whatever that test really does). > > The commit has no significant impact on any other test in the test suite. > > In that case I will certainly be interested how to reproduce that test. So I > need the following information: > > Machine description: Intel IvyBridge 2 sockets, 32 cores, 64G RAM > Config file: http://wherever.you.store/your/results/test-nr/config We have some information about this before. But not organized good enough, will improve it. > Test: http://wherever.you.store/your/tests/will-it-scale.per_process_ops.tar.bz2 > > That tarball should contain: > > README > test_script.sh > test_binary > > README should tell: > > will-it-scale.per_process_ops > > Short explanation of the test > > Preliminaries: > - perf > - whatever > > So that allows me to reproduce that test more or less with no effort. And > that's the really important part. For reproducing, now we use lkp-tests tool, which includes scripts to build the test case, run the test, collect various information, compare the test result, with the job file attached with the report email. That is not the easiest way, we will continuously improve it. > You can provide nice charts and full comparison tables for all tests on a web > site for those who are interested in large stats and pretty charts. > > Full results: http://wherever.you.store/your/results/test-nr/results Before we have a website for detailed information, we will still put some details into report email. > So now lets look at a scenario where that commit results in a performance > improvement on will-it-scale.per_process_ops and a regression on > will-it-scale.per_task. > > Subject: futex: 65d8fc777f6d: 25% performance regression > > FYI, we noticed a 25.2% performance regression due to commit > > 65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()" > > in the will-it-scale.per_task test. > > will-it-scale.per_tasks tests the futex operations for process private > futexes (Or whatever that test really does). > > The commit has significant impact on the following tests in the test suite: > > will-it-scale.per_process_ops: 25% improvement > > The 25% improvement is really not interesting. What's interesting is the > regression. That's what people need to look at. You still can provide the > information about all the tests including the 25% improvement data on > > Full results: http://wherever.you.store/your/results/test-nr/results > > Now, if you have commits which have mixed impact on various tests, then it's > important to point out the most relevant issue clearly in the > summary. I.e. you need to filter the tests for the maximum > regression/improvement and use that as the main information. You still can > provide the information about the impact on other tests in a very condensed > form. > > The commit has significant impact on the following tests in the test suite: > > test1: 0.5% Regression > test2: 2.0% Improvement > .... > testN: 0.2% Regression > > That's useful information, but it might be completely irrelevant if the > primary impact is, e.g. a 10% Regression. Then it's nice to know, that it gave > a 2% improvement on test2, but that's about it. The important information is > the 10% regression, which needs to be addressed urgently. Yes. We will separate summary and details. And put most important information at the front of of the summary. > Hope that helps. That really help us a lot! Thanks a lot! Best Regards, Huang, Ying > Thanks, > > tglx
[toc] | [prev] | [next] | [standalone]
| From | Thomas Gleixner <tglx@linutronix.de> |
|---|---|
| Date | 2016-03-21 09:10 +0100 |
| Subject | Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops |
| Message-ID | <rf34u-2NK-5@gated-at.bofh.it> |
| In reply to | #1361487 |
[Multipart message — attachments visible in raw view] — view raw
On Mon, 21 Mar 2016, Huang, Ying wrote:
> > FYI, we noticed 25.6% performance improvement due to commit
> >
> > 65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()"
> >
> > in the will-it-scale.per_process_ops test.
> >
> > will-it-scale.per_process_ops tests the futex operations for process shared
> > futexes (Or whatever that test really does).
>
> There is a futex sub test case for will-it-scale test suite. But I got your
> point, we need some description for the test case. If email is not too
> limited for the full description, we will put it in some web site and
> include short description and link to the full description in email.
Ok. Just make sure the short description gives enough information for the
casual reader.
> > The commit has no significant impact on any other test in the test suite.
>
> Sorry, we have no enough machine power to test all test cases for each
> bisect result. So we will have no such information until we find a way
> to do that.
Well, then I really have to ask how I should interpret the data here:
5076304 ± 0% +25.6% 6374220 ± 0% will-it-scale.per_process_ops
^^^ That's the reason why you sent the mail in the first place
1194117 ± 0% +14.4% 1366153 ± 1% will-it-scale.per_thread_ops
0.58 ± 0% -2.0% 0.57 ± 0% will-it-scale.scalability
6820 ± 0% -19.6% 5483 ± 15% meminfo.AnonHugePages
2652 ± 5% -10.4% 2375 ± 2% vmstat.system.cs
2848 ± 32% +141.2% 6870 ± 65% numa-meminfo.node1.Active(anon)
2832 ± 31% +57.6% 4465 ± 27% numa-meminfo.node1.AnonPages
15018 ± 12% -23.3% 11515 ± 15% numa-meminfo.node2.AnonPages
1214 ± 14% -22.8% 936.75 ± 20% numa-meminfo.node3.PageTables
712.25 ± 32% +141.2% 1718 ± 65% numa-vmstat.node1.nr_active_anon
708.25 ± 31% +57.7% 1116 ± 27% numa-vmstat.node1.nr_anon_pages
How is this related and what should I do about this information?
If it's important then I have to admit, that I fail to understand why.
If it's not important then I have to ask why is this included.
> > So that allows me to reproduce that test more or less with no effort. And
> > that's the really important part.
>
> For reproducing, now we use lkp-tests tool, which includes scripts to
> build the test case, run the test, collect various information, compare
> the test result, with the job file attached with the report email. That
> is not the easiest way, we will continuously improve it.
I know and lkp-tests is a pain to work with. So please look into a way to
extract the relevant binaries, so it's simple for developers to reproduce.
> > You can provide nice charts and full comparison tables for all tests on a web
> > site for those who are interested in large stats and pretty charts.
> >
> > Full results: http://wherever.you.store/your/results/test-nr/results
>
> Before we have a website for detailed information, we will still put
> some details into report email.
Ok, but please make them understandable for mere mortals.
Thanks,
tglx
[toc] | [prev] | [next] | [standalone]
| From | "Huang\, Ying" <ying.huang@intel.com> |
|---|---|
| Date | 2016-03-21 09:50 +0100 |
| Message-ID | <rf3Hb-337-9@gated-at.bofh.it> |
| In reply to | #1361570 |
Thomas Gleixner <tglx@linutronix.de> writes: > On Mon, 21 Mar 2016, Huang, Ying wrote: >> > FYI, we noticed 25.6% performance improvement due to commit >> > >> > 65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()" >> > >> > in the will-it-scale.per_process_ops test. >> > >> > will-it-scale.per_process_ops tests the futex operations for process shared >> > futexes (Or whatever that test really does). >> >> There is a futex sub test case for will-it-scale test suite. But I got your >> point, we need some description for the test case. If email is not too >> limited for the full description, we will put it in some web site and >> include short description and link to the full description in email. > > Ok. Just make sure the short description gives enough information for the > casual reader. > >> > The commit has no significant impact on any other test in the test suite. >> >> Sorry, we have no enough machine power to test all test cases for each >> bisect result. So we will have no such information until we find a way >> to do that. > > Well, then I really have to ask how I should interpret the data here: > > 5076304 ± 0% +25.6% 6374220 ± 0% will-it-scale.per_process_ops > > ^^^ That's the reason why you sent the mail in the first place > > 1194117 ± 0% +14.4% 1366153 ± 1% will-it-scale.per_thread_ops > 0.58 ± 0% -2.0% 0.57 ± 0% will-it-scale.scalability > 6820 ± 0% -19.6% 5483 ± 15% meminfo.AnonHugePages > 2652 ± 5% -10.4% 2375 ± 2% vmstat.system.cs > 2848 ± 32% +141.2% 6870 ± 65% numa-meminfo.node1.Active(anon) > 2832 ± 31% +57.6% 4465 ± 27% numa-meminfo.node1.AnonPages > 15018 ± 12% -23.3% 11515 ± 15% numa-meminfo.node2.AnonPages > 1214 ± 14% -22.8% 936.75 ± 20% numa-meminfo.node3.PageTables > 712.25 ± 32% +141.2% 1718 ± 65% numa-vmstat.node1.nr_active_anon > 708.25 ± 31% +57.7% 1116 ± 27% numa-vmstat.node1.nr_anon_pages > > How is this related and what should I do about this information? For each will-it-scale sub test case, it will be run in both process mode and thread mode, and task number will change from 1 to CPU number. will-it-scale.per_thread_ops shows thread mode main result. will-it-scale.scalability is calculated to measure how per_process_ops and per_thread_ops scaled along with the task number. These are default behavior of will-it-scale test suite. Others are monitors output. That is, other information collected during test. For example, meminfo is a monitor to sampling /proc/meminfo contents, AnonHugePages is a line in it. meminfo.AnonHugePages is for the average value of AnonHugePages line of /proc/meminfo. Similarly vmstat.system.cs is the average value of cs column of system column group of /usr/bin/vmstat. We hope these information are helpful for root causing the regression. > If it's important then I have to admit, that I fail to understand why. > > If it's not important then I have to ask why is this included. > >> > So that allows me to reproduce that test more or less with no effort. And >> > that's the really important part. >> >> For reproducing, now we use lkp-tests tool, which includes scripts to >> build the test case, run the test, collect various information, compare >> the test result, with the job file attached with the report email. That >> is not the easiest way, we will continuously improve it. > > I know and lkp-tests is a pain to work with. So please look into a way to > extract the relevant binaries, so it's simple for developers to reproduce. OK. We will try to improve on this side. But it is not an easy task for us to provided easy to use simple binaries. Do you think something like Docker image is easy to use? >> > You can provide nice charts and full comparison tables for all tests on a web >> > site for those who are interested in large stats and pretty charts. >> > >> > Full results: http://wherever.you.store/your/results/test-nr/results >> >> Before we have a website for detailed information, we will still put >> some details into report email. > > Ok, but please make them understandable for mere mortals. Sure. Best Regards, Huang, Ying > Thanks, > > tglx
[toc] | [prev] | [next] | [standalone]
| From | Darren Hart <dvhart@infradead.org> |
|---|---|
| Date | 2016-03-28 19:20 +0200 |
| Subject | Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops |
| Message-ID | <rhIZz-1I8-9@gated-at.bofh.it> |
| In reply to | #1361612 |
On Mon, Mar 21, 2016 at 04:42:47PM +0800, Huang, Ying wrote: > Thomas Gleixner <tglx@linutronix.de> writes: > > > On Mon, 21 Mar 2016, Huang, Ying wrote: > >> > FYI, we noticed 25.6% performance improvement due to commit > >> > > >> > 65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()" > >> > > >> > in the will-it-scale.per_process_ops test. > >> > > >> > will-it-scale.per_process_ops tests the futex operations for process shared > >> > futexes (Or whatever that test really does). > >> > >> There is a futex sub test case for will-it-scale test suite. But I got your > >> point, we need some description for the test case. If email is not too > >> limited for the full description, we will put it in some web site and > >> include short description and link to the full description in email. > > > > Ok. Just make sure the short description gives enough information for the > > casual reader. > > > >> > The commit has no significant impact on any other test in the test suite. > >> > >> Sorry, we have no enough machine power to test all test cases for each > >> bisect result. So we will have no such information until we find a way > >> to do that. > > > > Well, then I really have to ask how I should interpret the data here: > > > > 5076304 ± 0% +25.6% 6374220 ± 0% will-it-scale.per_process_ops > > > > ^^^ That's the reason why you sent the mail in the first place > > > > 1194117 ± 0% +14.4% 1366153 ± 1% will-it-scale.per_thread_ops > > 0.58 ± 0% -2.0% 0.57 ± 0% will-it-scale.scalability > > 6820 ± 0% -19.6% 5483 ± 15% meminfo.AnonHugePages > > 2652 ± 5% -10.4% 2375 ± 2% vmstat.system.cs > > 2848 ± 32% +141.2% 6870 ± 65% numa-meminfo.node1.Active(anon) > > 2832 ± 31% +57.6% 4465 ± 27% numa-meminfo.node1.AnonPages > > 15018 ± 12% -23.3% 11515 ± 15% numa-meminfo.node2.AnonPages > > 1214 ± 14% -22.8% 936.75 ± 20% numa-meminfo.node3.PageTables > > 712.25 ± 32% +141.2% 1718 ± 65% numa-vmstat.node1.nr_active_anon > > 708.25 ± 31% +57.7% 1116 ± 27% numa-vmstat.node1.nr_anon_pages > > > > How is this related and what should I do about this information? > > For each will-it-scale sub test case, it will be run in both process > mode and thread mode, and task number will change from 1 to CPU number. > will-it-scale.per_thread_ops shows thread mode main result. > will-it-scale.scalability is calculated to measure how per_process_ops > and per_thread_ops scaled along with the task number. These are default > behavior of will-it-scale test suite. > > Others are monitors output. That is, other information collected during > test. For example, meminfo is a monitor to sampling /proc/meminfo > contents, AnonHugePages is a line in it. meminfo.AnonHugePages is for > the average value of AnonHugePages line of /proc/meminfo. Similarly > vmstat.system.cs is the average value of cs column of system column > group of /usr/bin/vmstat. > > We hope these information are helpful for root causing the regression. > > > If it's important then I have to admit, that I fail to understand why. > > > > If it's not important then I have to ask why is this included. > > > >> > So that allows me to reproduce that test more or less with no effort. And > >> > that's the really important part. > >> > >> For reproducing, now we use lkp-tests tool, which includes scripts to > >> build the test case, run the test, collect various information, compare > >> the test result, with the job file attached with the report email. That > >> is not the easiest way, we will continuously improve it. > > > > I know and lkp-tests is a pain to work with. So please look into a way to > > extract the relevant binaries, so it's simple for developers to reproduce. > > OK. We will try to improve on this side. But it is not an easy task > for us to provided easy to use simple binaries. Do you think something > like Docker image is easy to use? Thomas, I presume you are interested in binaries to be positive we're reproducing with exactly the same bits? I agree that's good to have. I'd also want to have or be pointed to the sources with a straight forward way to rebuild and to inspect what exactly the test is doing. (I assume this is implied, but just to make sure it's stated). Huang, what makes the binaries difficult to package? And how would docker make that any simpler? > > >> > You can provide nice charts and full comparison tables for all tests on a web > >> > site for those who are interested in large stats and pretty charts. > >> > > >> > Full results: http://wherever.you.store/your/results/test-nr/results > >> > >> Before we have a website for detailed information, we will still put > >> some details into report email. > > > > Ok, but please make them understandable for mere mortals. Thanks for speaking up for us mortals Thomas ;-) -- Darren Hart Intel Open Source Technology Center
[toc] | [prev] | [next] | [standalone]
| From | "Huang\, Ying" <ying.huang@intel.com> |
|---|---|
| Date | 2016-03-29 03:20 +0200 |
| Message-ID | <rhQu6-77q-13@gated-at.bofh.it> |
| In reply to | #1365592 |
Darren Hart <dvhart@infradead.org> writes: > On Mon, Mar 21, 2016 at 04:42:47PM +0800, Huang, Ying wrote: >> Thomas Gleixner <tglx@linutronix.de> writes: >> >> > On Mon, 21 Mar 2016, Huang, Ying wrote: >> >> > FYI, we noticed 25.6% performance improvement due to commit >> >> > >> >> > 65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()" >> >> > >> >> > in the will-it-scale.per_process_ops test. >> >> > >> >> > will-it-scale.per_process_ops tests the futex operations for process shared >> >> > futexes (Or whatever that test really does). >> >> >> >> There is a futex sub test case for will-it-scale test suite. But I got your >> >> point, we need some description for the test case. If email is not too >> >> limited for the full description, we will put it in some web site and >> >> include short description and link to the full description in email. >> > >> > Ok. Just make sure the short description gives enough information for the >> > casual reader. >> > >> >> > The commit has no significant impact on any other test in the test suite. >> >> >> >> Sorry, we have no enough machine power to test all test cases for each >> >> bisect result. So we will have no such information until we find a way >> >> to do that. >> > >> > Well, then I really have to ask how I should interpret the data here: >> > >> > 5076304 . 0% +25.6% 6374220 . 0% will-it-scale.per_process_ops >> > >> > ^^^ That's the reason why you sent the mail in the first place >> > >> > 1194117 . 0% +14.4% 1366153 . 1% will-it-scale.per_thread_ops >> > 0.58 . 0% -2.0% 0.57 . 0% will-it-scale.scalability >> > 6820 . 0% -19.6% 5483 . 15% meminfo.AnonHugePages >> > 2652 . 5% -10.4% 2375 . 2% vmstat.system.cs >> > 2848 . 32% +141.2% 6870 . 65% numa-meminfo.node1.Active(anon) >> > 2832 . 31% +57.6% 4465 . 27% numa-meminfo.node1.AnonPages >> > 15018 . 12% -23.3% 11515 . 15% numa-meminfo.node2.AnonPages >> > 1214 . 14% -22.8% 936.75 . 20% numa-meminfo.node3.PageTables >> > 712.25 . 32% +141.2% 1718 . 65% numa-vmstat.node1.nr_active_anon >> > 708.25 . 31% +57.7% 1116 . 27% numa-vmstat.node1.nr_anon_pages >> > >> > How is this related and what should I do about this information? >> >> For each will-it-scale sub test case, it will be run in both process >> mode and thread mode, and task number will change from 1 to CPU number. >> will-it-scale.per_thread_ops shows thread mode main result. >> will-it-scale.scalability is calculated to measure how per_process_ops >> and per_thread_ops scaled along with the task number. These are default >> behavior of will-it-scale test suite. >> >> Others are monitors output. That is, other information collected during >> test. For example, meminfo is a monitor to sampling /proc/meminfo >> contents, AnonHugePages is a line in it. meminfo.AnonHugePages is for >> the average value of AnonHugePages line of /proc/meminfo. Similarly >> vmstat.system.cs is the average value of cs column of system column >> group of /usr/bin/vmstat. >> >> We hope these information are helpful for root causing the regression. >> >> > If it's important then I have to admit, that I fail to understand why. >> > >> > If it's not important then I have to ask why is this included. >> > >> >> > So that allows me to reproduce that test more or less with no effort. And >> >> > that's the really important part. >> >> >> >> For reproducing, now we use lkp-tests tool, which includes scripts to >> >> build the test case, run the test, collect various information, compare >> >> the test result, with the job file attached with the report email. That >> >> is not the easiest way, we will continuously improve it. >> > >> > I know and lkp-tests is a pain to work with. So please look into a way to >> > extract the relevant binaries, so it's simple for developers to reproduce. >> >> OK. We will try to improve on this side. But it is not an easy task >> for us to provided easy to use simple binaries. Do you think something >> like Docker image is easy to use? > > Thomas, I presume you are interested in binaries to be positive we're > reproducing with exactly the same bits? I agree that's good to have. I'd also > want to have or be pointed to the sources with a straight forward way to > rebuild and to inspect what exactly the test is doing. (I assume this is > implied, but just to make sure it's stated). lkp-tests has the scripts to download the source, apply some patch, and build the binary. It is not a very straight forward way, but the script is quite simple. > Huang, what makes the binaries difficult to package? And how would docker make > that any simpler? The binaries is not difficult to package. But the test is not only benchmark binary. We may do some setup for the specific test, for example, change the cpu frequency governor, mkfs on a partition (may be ramdisk), etc. And we have mechanism to collect various information during the test, for example, run vmstat, sampling /proc/sched_debug, etc. As for docker, we just want to reduce the pain of using lkp-tests. >> >> > You can provide nice charts and full comparison tables for all tests on a web >> >> > site for those who are interested in large stats and pretty charts. >> >> > >> >> > Full results: http://wherever.you.store/your/results/test-nr/results >> >> >> >> Before we have a website for detailed information, we will still put >> >> some details into report email. >> > >> > Ok, but please make them understandable for mere mortals. > > Thanks for speaking up for us mortals Thomas ;-) Best Regards, Huang, Ying
[toc] | [prev] | [next] | [standalone]
| From | Darren Hart <dvhart@infradead.org> |
|---|---|
| Date | 2016-03-29 07:20 +0200 |
| Subject | Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops |
| Message-ID | <rhUel-1qg-1@gated-at.bofh.it> |
| In reply to | #1365754 |
On Tue, Mar 29, 2016 at 09:12:56AM +0800, Huang, Ying wrote: > Darren Hart <dvhart@infradead.org> writes: > > > On Mon, Mar 21, 2016 at 04:42:47PM +0800, Huang, Ying wrote: > >> Thomas Gleixner <tglx@linutronix.de> writes: > >> > >> > On Mon, 21 Mar 2016, Huang, Ying wrote: > >> >> > FYI, we noticed 25.6% performance improvement due to commit > >> >> > > >> >> > 65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()" > >> >> > > >> >> > in the will-it-scale.per_process_ops test. > >> >> > > >> >> > will-it-scale.per_process_ops tests the futex operations for process shared > >> >> > futexes (Or whatever that test really does). > >> >> > >> >> There is a futex sub test case for will-it-scale test suite. But I got your > >> >> point, we need some description for the test case. If email is not too > >> >> limited for the full description, we will put it in some web site and > >> >> include short description and link to the full description in email. > >> > > >> > Ok. Just make sure the short description gives enough information for the > >> > casual reader. > >> > > >> >> > The commit has no significant impact on any other test in the test suite. > >> >> > >> >> Sorry, we have no enough machine power to test all test cases for each > >> >> bisect result. So we will have no such information until we find a way > >> >> to do that. > >> > > >> > Well, then I really have to ask how I should interpret the data here: > >> > > >> > 5076304 . 0% +25.6% 6374220 . 0% will-it-scale.per_process_ops > >> > > >> > ^^^ That's the reason why you sent the mail in the first place > >> > > >> > 1194117 . 0% +14.4% 1366153 . 1% will-it-scale.per_thread_ops > >> > 0.58 . 0% -2.0% 0.57 . 0% will-it-scale.scalability > >> > 6820 . 0% -19.6% 5483 . 15% meminfo.AnonHugePages > >> > 2652 . 5% -10.4% 2375 . 2% vmstat.system.cs > >> > 2848 . 32% +141.2% 6870 . 65% numa-meminfo.node1.Active(anon) > >> > 2832 . 31% +57.6% 4465 . 27% numa-meminfo.node1.AnonPages > >> > 15018 . 12% -23.3% 11515 . 15% numa-meminfo.node2.AnonPages > >> > 1214 . 14% -22.8% 936.75 . 20% numa-meminfo.node3.PageTables > >> > 712.25 . 32% +141.2% 1718 . 65% numa-vmstat.node1.nr_active_anon > >> > 708.25 . 31% +57.7% 1116 . 27% numa-vmstat.node1.nr_anon_pages > >> > > >> > How is this related and what should I do about this information? > >> > >> For each will-it-scale sub test case, it will be run in both process > >> mode and thread mode, and task number will change from 1 to CPU number. > >> will-it-scale.per_thread_ops shows thread mode main result. > >> will-it-scale.scalability is calculated to measure how per_process_ops > >> and per_thread_ops scaled along with the task number. These are default > >> behavior of will-it-scale test suite. > >> > >> Others are monitors output. That is, other information collected during > >> test. For example, meminfo is a monitor to sampling /proc/meminfo > >> contents, AnonHugePages is a line in it. meminfo.AnonHugePages is for > >> the average value of AnonHugePages line of /proc/meminfo. Similarly > >> vmstat.system.cs is the average value of cs column of system column > >> group of /usr/bin/vmstat. > >> > >> We hope these information are helpful for root causing the regression. > >> > >> > If it's important then I have to admit, that I fail to understand why. > >> > > >> > If it's not important then I have to ask why is this included. > >> > > >> >> > So that allows me to reproduce that test more or less with no effort. And > >> >> > that's the really important part. > >> >> > >> >> For reproducing, now we use lkp-tests tool, which includes scripts to > >> >> build the test case, run the test, collect various information, compare > >> >> the test result, with the job file attached with the report email. That > >> >> is not the easiest way, we will continuously improve it. > >> > > >> > I know and lkp-tests is a pain to work with. So please look into a way to > >> > extract the relevant binaries, so it's simple for developers to reproduce. > >> > >> OK. We will try to improve on this side. But it is not an easy task > >> for us to provided easy to use simple binaries. Do you think something > >> like Docker image is easy to use? > > > > Thomas, I presume you are interested in binaries to be positive we're > > reproducing with exactly the same bits? I agree that's good to have. I'd also > > want to have or be pointed to the sources with a straight forward way to > > rebuild and to inspect what exactly the test is doing. (I assume this is > > implied, but just to make sure it's stated). > > lkp-tests has the scripts to download the source, apply some patch, and > build the binary. It is not a very straight forward way, but the script > is quite simple. > > > Huang, what makes the binaries difficult to package? And how would docker make > > that any simpler? > > The binaries is not difficult to package. But the test is not only > benchmark binary. We may do some setup for the specific test, for > example, change the cpu frequency governor, mkfs on a partition (may be > ramdisk), etc. And we have mechanism to collect various information > during the test, for example, run vmstat, sampling /proc/sched_debug, > etc. In that case, it would be really useful to document your setup and include a link to that in every report you send out. It should include things like the partition layout (which ideally would have normalized symlinks to facilitate reproduction outside your lab) (/dev/testpart -> /dev/sde3) and the scripts/tests should only use the symlinks. That's an over simplification of course, but that kind of configuration and documentation would be very helpful in reducing the barrier to getting people to look at the issues your testing discovers - and that, of course, is the whole point. Thank you for doing the work and accepting all this feedback. > As for docker, we just want to reduce the pain of using lkp-tests. > Sometimes docker is pain as well. It became completely unusable on my Debian 8 system not too long ago. That can become another barrier if it isn't also documented without docker, even if it requires more steps. -- Darren Hart Intel Open Source Technology Center
[toc] | [prev] | [next] | [standalone]
| From | "Huang\, Ying" <ying.huang@intel.com> |
|---|---|
| Date | 2016-03-29 08:00 +0200 |
| Message-ID | <rhUR3-1DN-1@gated-at.bofh.it> |
| In reply to | #1365856 |
Darren Hart <dvhart@infradead.org> writes: > On Tue, Mar 29, 2016 at 09:12:56AM +0800, Huang, Ying wrote: >> Darren Hart <dvhart@infradead.org> writes: >> >> > On Mon, Mar 21, 2016 at 04:42:47PM +0800, Huang, Ying wrote: >> >> Thomas Gleixner <tglx@linutronix.de> writes: >> >> >> >> > On Mon, 21 Mar 2016, Huang, Ying wrote: >> >> >> > FYI, we noticed 25.6% performance improvement due to commit >> >> >> > >> >> >> > 65d8fc777f6d "futex: Remove requirement for lock_page() in get_futex_key()" >> >> >> > >> >> >> > in the will-it-scale.per_process_ops test. >> >> >> > >> >> >> > will-it-scale.per_process_ops tests the futex operations for process shared >> >> >> > futexes (Or whatever that test really does). >> >> >> >> >> >> There is a futex sub test case for will-it-scale test suite. But I got your >> >> >> point, we need some description for the test case. If email is not too >> >> >> limited for the full description, we will put it in some web site and >> >> >> include short description and link to the full description in email. >> >> > >> >> > Ok. Just make sure the short description gives enough information for the >> >> > casual reader. >> >> > >> >> >> > The commit has no significant impact on any other test in the test suite. >> >> >> >> >> >> Sorry, we have no enough machine power to test all test cases for each >> >> >> bisect result. So we will have no such information until we find a way >> >> >> to do that. >> >> > >> >> > Well, then I really have to ask how I should interpret the data here: >> >> > >> >> > 5076304 . 0% +25.6% 6374220 . 0% will-it-scale.per_process_ops >> >> > >> >> > ^^^ That's the reason why you sent the mail in the first place >> >> > >> >> > 1194117 . 0% +14.4% 1366153 . 1% will-it-scale.per_thread_ops >> >> > 0.58 . 0% -2.0% 0.57 . 0% will-it-scale.scalability >> >> > 6820 . 0% -19.6% 5483 . 15% meminfo.AnonHugePages >> >> > 2652 . 5% -10.4% 2375 . 2% vmstat.system.cs >> >> > 2848 . 32% +141.2% 6870 . 65% numa-meminfo.node1.Active(anon) >> >> > 2832 . 31% +57.6% 4465 . 27% numa-meminfo.node1.AnonPages >> >> > 15018 . 12% -23.3% 11515 . 15% numa-meminfo.node2.AnonPages >> >> > 1214 . 14% -22.8% 936.75 . 20% numa-meminfo.node3.PageTables >> >> > 712.25 . 32% +141.2% 1718 . 65% numa-vmstat.node1.nr_active_anon >> >> > 708.25 . 31% +57.7% 1116 . 27% numa-vmstat.node1.nr_anon_pages >> >> > >> >> > How is this related and what should I do about this information? >> >> >> >> For each will-it-scale sub test case, it will be run in both process >> >> mode and thread mode, and task number will change from 1 to CPU number. >> >> will-it-scale.per_thread_ops shows thread mode main result. >> >> will-it-scale.scalability is calculated to measure how per_process_ops >> >> and per_thread_ops scaled along with the task number. These are default >> >> behavior of will-it-scale test suite. >> >> >> >> Others are monitors output. That is, other information collected during >> >> test. For example, meminfo is a monitor to sampling /proc/meminfo >> >> contents, AnonHugePages is a line in it. meminfo.AnonHugePages is for >> >> the average value of AnonHugePages line of /proc/meminfo. Similarly >> >> vmstat.system.cs is the average value of cs column of system column >> >> group of /usr/bin/vmstat. >> >> >> >> We hope these information are helpful for root causing the regression. >> >> >> >> > If it's important then I have to admit, that I fail to understand why. >> >> > >> >> > If it's not important then I have to ask why is this included. >> >> > >> >> >> > So that allows me to reproduce that test more or less with no effort. And >> >> >> > that's the really important part. >> >> >> >> >> >> For reproducing, now we use lkp-tests tool, which includes scripts to >> >> >> build the test case, run the test, collect various information, compare >> >> >> the test result, with the job file attached with the report email. That >> >> >> is not the easiest way, we will continuously improve it. >> >> > >> >> > I know and lkp-tests is a pain to work with. So please look into a way to >> >> > extract the relevant binaries, so it's simple for developers to reproduce. >> >> >> >> OK. We will try to improve on this side. But it is not an easy task >> >> for us to provided easy to use simple binaries. Do you think something >> >> like Docker image is easy to use? >> > >> > Thomas, I presume you are interested in binaries to be positive we're >> > reproducing with exactly the same bits? I agree that's good to have. I'd also >> > want to have or be pointed to the sources with a straight forward way to >> > rebuild and to inspect what exactly the test is doing. (I assume this is >> > implied, but just to make sure it's stated). >> >> lkp-tests has the scripts to download the source, apply some patch, and >> build the binary. It is not a very straight forward way, but the script >> is quite simple. >> >> > Huang, what makes the binaries difficult to package? And how would docker make >> > that any simpler? >> >> The binaries is not difficult to package. But the test is not only >> benchmark binary. We may do some setup for the specific test, for >> example, change the cpu frequency governor, mkfs on a partition (may be >> ramdisk), etc. And we have mechanism to collect various information >> during the test, for example, run vmstat, sampling /proc/sched_debug, >> etc. > > In that case, it would be really useful to document your setup and include a > link to that in every report you send out. The setup information is in the job file attached in the report email. The job file can be used together with lkp-tests to reproduce exactly we done in our test machine. > It should include things like the > partition layout (which ideally would have normalized symlinks to facilitate > reproduction outside your lab) (/dev/testpart -> /dev/sde3) and the > scripts/tests should only use the symlinks. That's an over simplification of > course, but that kind of configuration and documentation would be very helpful > in reducing the barrier to getting people to look at the issues your testing > discovers - and that, of course, is the whole point. The test partition of your test machine can be specified in the <host> file in lkp-tests. Our current reproducing solution is based on lkp-tests. We hope we can improve the lkp-tests tool to make reproduction easier. > Thank you for doing the work and accepting all this feedback. Our pleasure. >> As for docker, we just want to reduce the pain of using lkp-tests. >> > > Sometimes docker is pain as well. It became completely unusable on my Debian 8 > system not too long ago. That can become another barrier if it isn't also > documented without docker, even if it requires more steps. OK. I see. Your input are valuable for us. Best Regards, Huang, Ying
[toc] | [prev] | [next] | [standalone]
| From | Peter Zijlstra <peterz@infradead.org> |
|---|---|
| Date | 2016-03-29 09:50 +0200 |
| Subject | Re: [LKP] [lkp] [futex] 65d8fc777f: +25.6% will-it-scale.per_process_ops |
| Message-ID | <rhWzv-2QS-1@gated-at.bofh.it> |
| In reply to | #1361612 |
On Mon, Mar 21, 2016 at 04:42:47PM +0800, Huang, Ying wrote: > OK. We will try to improve on this side. But it is not an easy task > for us to provided easy to use simple binaries. Do you think something > like Docker image is easy to use? Docker muck would just make me turn around and run.
[toc] | [prev] | [standalone]
Back to top | Article view | linux.kernel
csiph-web