[<prev] [next>] [day] [month] [year] [list]
Message-ID: <20170609052939.GA21477@yexl-desktop>
Date: Fri, 9 Jun 2017 13:29:39 +0800
From: kernel test robot <xiaolong.ye@...el.com>
To: Dave Airlie <airlied@...hat.com>
Cc: Sean Paul <seanpaul@...omium.org>,
LKML <linux-kernel@...r.kernel.org>,
Dave Airlie <airlied@...hat.com>,
dri-devel@...ts.freedesktop.org, lkp@...org
Subject: [lkp-robot] [drm] e1f8a89c4f: pft.faults_per_sec_per_cpu -77.9%
regression
Greeting,
FYI, we noticed a -77.9% regression of pft.faults_per_sec_per_cpu due to commit:
commit: e1f8a89c4f33f5b32cfb047f07fdf6d38cda854a ("drm: introduce sync objects (v4)")
git://people.freedesktop.org/~airlied/linux.git drm-syncobj-sem
in testcase: pft
on test machine: qemu-system-x86_64 -enable-kvm -cpu Penryn -smp 4 -m 2G
with following parameters:
iterations: 20x
test-description: Pft is the page fault test micro benchmark.
test-url: https://github.com/gormanm/pft
Details are as below:
-------------------------------------------------------------------------------------------------->
To reproduce:
git clone https://github.com/01org/lkp-tests.git
cd lkp-tests
bin/lkp qemu -k <bzImage> job-script # job-script is attached in this email
testcase/path_params/tbox_group/run: pft/20x/vm-vp-2G
v4.12-rc2 e1f8a89c4f33f5b32cfb047f07
---------------- --------------------------
%stddev %change %stddev \ | \ 878620 ± 15% -77.9% 194209 ± 12% pft.faults_per_sec_per_cpu
7.16 ± 23% +257.2% 25.58 ± 12% pft.time.elapsed_time
7.16 ± 23% +257.2% 25.58 ± 12% pft.time.elapsed_time.max
1223 ± 10% +199.9% 3667 ± 15% pft.time.involuntary_context_switches
61828 ± 2% +10106.0% 6310202 ± 0% pft.time.minor_page_faults
145.40 ± 3% +20.6% 175.33 ± 5% pft.time.percent_of_cpu_this_job_got
9.37 ± 22% +345.2% 41.72 ± 15% pft.time.system_time
5.04 ± 11% -45.6% 2.74 ± 9% mpstat.cpu.usr%
131734 ± 12% +20.3% 158487 ± 5% softirqs.TIMER 1335 ± 11% +105.1% 2738 ± 19% vmstat.system.cs
5390 ± 4% -27.6% 3904 ± 38% slabinfo.kmalloc-96.active_objs
5420 ± 3% -21.5% 4252 ± 27% slabinfo.kmalloc-96.num_objs
68.30 ±209% +13218.2% 9096 ± 84% latency_stats.max.pipe_read.__vfs_read.vfs_read.SyS_read.entry_SYSCALL_64_fastpath
588552 ± 32% -74.2% 151873 ± 37% latency_stats.sum.ep_poll.SyS_epoll_wait.do_syscall_64.return_from_SYSCALL_64
74.20 ±203% +12326.8% 9220 ± 82% latency_stats.sum.pipe_read.__vfs_read.vfs_read.SyS_read.entry_SYSCALL_64_fastpath
160550 ± 14% -100.0% 0.00 ± -1% meminfo.AnonHugePages
204792 ± 0% -100.0% 0.00 ± -1% meminfo.CmaFree
204800 ± 0% -100.0% 0.00 ± -1% meminfo.CmaTotal
3041 ± 0% +13.4% 3449 ± 0% meminfo.KernelStack
51198 ± 0% -100.0% 0.00 ± -1% proc-vmstat.nr_free_cma
3033 ± 1% +13.6% 3446 ± 0% proc-vmstat.nr_kernel_stack
73394 ± 9% +8258.9% 6134982 ± 2% proc-vmstat.numa_hit
73394 ± 9% +8258.9% 6134982 ± 2% proc-vmstat.numa_local
67006 ± 9% +9043.7% 6126912 ± 2% proc-vmstat.pgfault
11161 ± 7% -100.0% 0.00 ± -1% proc-vmstat.thp_deferred_split_page
11217 ± 7% -100.0% 0.00 ± -1% proc-vmstat.thp_fault_alloc
7.16 ± 23% +257.2% 25.58 ± 12% time.elapsed_time
7.16 ± 23% +257.2% 25.58 ± 12% time.elapsed_time.max
1223 ± 10% +199.9% 3667 ± 15% time.involuntary_context_switches
61828 ± 2% +10106.0% 6310202 ± 0% time.minor_page_faults
pft.faults_per_sec_per_cpu
1e+06 ++--------------------------------------*--------------------------+
900000 ++ *.. .. |
| * : .. |
800000 ++ : : : ...* |
700000 ++ ...* : : : *... |
*... : : : : |
600000 ++ : : : : |
500000 ++ : : : : |
400000 ++ : : : : |
| : : : : |
300000 ++ : : : : |
200000 O+ O : O : O : O : O O O
| : : : : O O |
100000 ++ : : : : |
0 ++-----------*-------------*-------------------O-------------------+
[*] bisect-good sample
[O] bisect-bad sample
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,
Xiaolong
View attachment "config-4.12.0-rc2-00001-ge1f8a89" of type "text/plain" (117141 bytes)
View attachment "job-script" of type "text/plain" (6280 bytes)
Powered by blists - more mailing lists