FYI, we noticed netperf.Throughput_Mbps 14.2% improvement due to commit:
commit 2159197d66770ec01f75c93fb11dc66df81fd45b ("sched/core: Enable increased
load resolution on 64-bit kernels")
https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git sched/core
in testcase - netperf
on test machine - lkp-hsw-d01: 8 threads Haswell with 8G memory
=========================================================================================
cluster/compiler/cpufreq_governor/ip/kconfig/nr_threads/rootfs/runtime/send_size/tbox_group/test/testcase:
cs-localhost/gcc-4.9/performance/ipv4/x86_64-rhel/200%/debian-x86_64-2015-02-07.cgz/300s/10K/lkp-hsw-d01/SCTP_STREAM_MANY/netperf
commit:
e7904a28f5331c21d17af638cb477c83662e3cb6
2159197d66770ec01f75c93fb11dc66df81fd45b
e7904a28f5331c21 2159197d66770ec01f75c93fb1
---------------- --------------------------
%stddev %change %stddev
\ | \
4657 ± 3% +14.2% 5317 ± 0% netperf.Throughput_Mbps
10987 ± 9% +3.4e+05% 37810767 ± 0%
netperf.time.involuntary_context_switches
527.25 ± 0% -25.1% 395.00 ± 0%
netperf.time.percent_of_cpu_this_job_got
1547 ± 0% -24.9% 1162 ± 0% netperf.time.system_time
44.45 ± 1% -32.1% 30.16 ± 1% netperf.time.user_time
16045063 ± 3% -82.8% 2757566 ± 6%
netperf.time.voluntary_context_switches
10119 ± 41% +203.3% 30688 ± 38% cpuidle.C1-HSW.usage
51528 ± 8% +203.0% 156148 ± 9% softirqs.RCU
55.94 ± 1% +4.2% 58.31 ± 0% turbostat.CorWatt
64.79 ± 1% +2.6% 66.45 ± 0% turbostat.PkgWatt
21.00 ± 0% -8.3% 19.25 ± 2% vmstat.procs.r
105875 ± 3% +155.5% 270467 ± 0% vmstat.system.cs
3.535e+08 ± 3% +13.4% 4.007e+08 ± 0% proc-vmstat.numa_hit
3.535e+08 ± 3% +13.4% 4.007e+08 ± 0% proc-vmstat.numa_local
8.196e+08 ± 3% +22.1% 1.001e+09 ± 0% proc-vmstat.pgalloc_dma32
1.043e+09 ± 3% +23.0% 1.282e+09 ± 0% proc-vmstat.pgalloc_normal
1.862e+09 ± 3% +22.6% 2.283e+09 ± 0% proc-vmstat.pgfree
10987 ± 9% +3.4e+05% 37810767 ± 0% time.involuntary_context_switches
527.25 ± 0% -25.1% 395.00 ± 0% time.percent_of_cpu_this_job_got
1547 ± 0% -24.9% 1162 ± 0% time.system_time
44.45 ± 1% -32.1% 30.16 ± 1% time.user_time
16045063 ± 3% -82.8% 2757566 ± 6% time.voluntary_context_switches
727242 ± 6% -94.3% 41723 ± 17%
sched_debug.cfs_rq:/.MIN_vruntime.avg
824178 ± 0% -90.6% 77484 ± 0%
sched_debug.cfs_rq:/.MIN_vruntime.max
353474 ± 28% -100.0% 0.00 ± 0%
sched_debug.cfs_rq:/.MIN_vruntime.min
181141 ± 29% -80.4% 35590 ± 4%
sched_debug.cfs_rq:/.MIN_vruntime.stddev
215.56 ± 4% +6.5e+05% 1404265 ± 3% sched_debug.cfs_rq:/.load.avg
289.96 ± 27% +6.7e+05% 1954288 ± 3% sched_debug.cfs_rq:/.load.max
161.25 ± 6% +5.4e+05% 865368 ± 0% sched_debug.cfs_rq:/.load.min
41.64 ± 61% +1.2e+06% 503455 ± 3% sched_debug.cfs_rq:/.load.stddev
196.06 ± 2% +401.4% 983.07 ± 2% sched_debug.cfs_rq:/.load_avg.avg
241.33 ± 7% +359.5% 1108 ± 7% sched_debug.cfs_rq:/.load_avg.max
161.29 ± 0% +434.5% 862.17 ± 0% sched_debug.cfs_rq:/.load_avg.min
31.50 ± 19% +180.5% 88.36 ± 34%
sched_debug.cfs_rq:/.load_avg.stddev
727242 ± 6% -94.3% 41723 ± 17%
sched_debug.cfs_rq:/.max_vruntime.avg
824178 ± 0% -90.6% 77484 ± 0%
sched_debug.cfs_rq:/.max_vruntime.max
353474 ± 28% -100.0% 0.00 ± 0%
sched_debug.cfs_rq:/.max_vruntime.min
181141 ± 29% -80.4% 35590 ± 4%
sched_debug.cfs_rq:/.max_vruntime.stddev
819702 ± 0% -90.7% 76071 ± 0%
sched_debug.cfs_rq:/.min_vruntime.avg
826026 ± 0% -90.4% 79401 ± 1%
sched_debug.cfs_rq:/.min_vruntime.max
815229 ± 0% -90.8% 74929 ± 0%
sched_debug.cfs_rq:/.min_vruntime.min
3308 ± 26% -57.2% 1415 ± 23%
sched_debug.cfs_rq:/.min_vruntime.stddev
1.66 ± 1% -18.9% 1.34 ± 3%
sched_debug.cfs_rq:/.nr_running.avg
1.25 ± 6% -33.3% 0.83 ± 0%
sched_debug.cfs_rq:/.nr_running.min
0.24 ± 18% +98.5% 0.48 ± 4%
sched_debug.cfs_rq:/.nr_running.stddev
151.14 ± 1% +482.7% 880.75 ± 1%
sched_debug.cfs_rq:/.runnable_load_avg.avg
168.83 ± 1% +502.6% 1017 ± 0%
sched_debug.cfs_rq:/.runnable_load_avg.max
103.42 ± 9% +601.1% 725.04 ± 10%
sched_debug.cfs_rq:/.runnable_load_avg.min
21.35 ± 17% +357.0% 97.56 ± 20%
sched_debug.cfs_rq:/.runnable_load_avg.stddev
-9567 ±-36% -53.2% -4474 ±-24% sched_debug.cfs_rq:/.spread0.min
3310 ± 26% -57.2% 1416 ± 23%
sched_debug.cfs_rq:/.spread0.stddev
536142 ± 16% -39.3% 325631 ± 29% sched_debug.cpu.avg_idle.avg
0.90 ± 14% +150.7% 2.26 ± 5% sched_debug.cpu.clock.stddev
0.90 ± 14% +150.7% 2.26 ± 5% sched_debug.cpu.clock_task.stddev
153.77 ± 0% +477.9% 888.63 ± 1% sched_debug.cpu.cpu_load[0].avg
169.17 ± 1% +503.3% 1020 ± 0% sched_debug.cpu.cpu_load[0].max
122.96 ± 5% +549.6% 798.71 ± 1% sched_debug.cpu.cpu_load[0].min
14.98 ± 13% +408.1% 76.13 ± 2% sched_debug.cpu.cpu_load[0].stddev
153.47 ± 0% +484.2% 896.64 ± 1% sched_debug.cpu.cpu_load[1].avg
166.75 ± 0% +502.6% 1004 ± 1% sched_debug.cpu.cpu_load[1].max
130.62 ± 2% +514.7% 802.92 ± 3% sched_debug.cpu.cpu_load[1].min
11.46 ± 8% +483.7% 66.91 ± 10% sched_debug.cpu.cpu_load[1].stddev
152.94 ± 0% +485.7% 895.80 ± 1% sched_debug.cpu.cpu_load[2].avg
165.04 ± 0% +502.6% 994.50 ± 1% sched_debug.cpu.cpu_load[2].max
136.29 ± 1% +489.4% 803.29 ± 3% sched_debug.cpu.cpu_load[2].min
9.17 ± 12% +589.1% 63.21 ± 12% sched_debug.cpu.cpu_load[2].stddev
153.46 ± 0% +483.7% 895.78 ± 1% sched_debug.cpu.cpu_load[3].avg
163.92 ± 0% +500.3% 984.04 ± 1% sched_debug.cpu.cpu_load[3].max
141.88 ± 1% +468.7% 806.79 ± 2% sched_debug.cpu.cpu_load[3].min
7.18 ± 12% +734.9% 59.98 ± 12% sched_debug.cpu.cpu_load[3].stddev
155.51 ± 0% +475.5% 894.93 ± 1% sched_debug.cpu.cpu_load[4].avg
164.58 ± 0% +489.0% 969.38 ± 0% sched_debug.cpu.cpu_load[4].max
146.58 ± 1% +452.8% 810.38 ± 2% sched_debug.cpu.cpu_load[4].min
5.76 ± 13% +858.1% 55.18 ± 12% sched_debug.cpu.cpu_load[4].stddev
215.88 ± 5% +6.3e+05% 1365964 ± 6% sched_debug.cpu.load.avg
293.25 ± 26% +6.7e+05% 1953269 ± 3% sched_debug.cpu.load.max
166.50 ± 5% +5.2e+05% 865491 ± 0% sched_debug.cpu.load.min
40.05 ± 61% +1.2e+06% 478086 ± 4% sched_debug.cpu.load.stddev
2.51 ± 1% -10.6% 2.24 ± 3% sched_debug.cpu.nr_running.avg
1.96 ± 7% -27.7% 1.42 ± 13% sched_debug.cpu.nr_running.min
0.38 ± 24% +85.2% 0.70 ± 23% sched_debug.cpu.nr_running.stddev
2001135 ± 3% +154.6% 5095769 ± 0% sched_debug.cpu.nr_switches.avg
2026432 ± 4% +158.1% 5231149 ± 0% sched_debug.cpu.nr_switches.max
1975801 ± 3% +152.2% 4983341 ± 1% sched_debug.cpu.nr_switches.min
17661 ± 48% +365.1% 82150 ± 34% sched_debug.cpu.nr_switches.stddev
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,
Xiaolong
---
LKP_SERVER: inn
LKP_CGI_PORT: 80
LKP_CIFS_PORT: 139
testcase: netperf
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
NFS_HANG_DF_TIMEOUT: 200
NFS_HANG_CHECK_INTERVAL: 900
default-watchdogs:
oom-killer:
watchdog:
nfs-hang:
commit: 2159197d66770ec01f75c93fb11dc66df81fd45b
model: Haswell
nr_cpu: 8
memory: 8G
hdd_partitions:
swap_partitions:
rootfs_partition:
category: benchmark
disable_latency_stats: 1
ip: ipv4
runtime: 300s
nr_threads: 200%
perf-profile:
freq: 800
cluster: cs-localhost
if role server:
netserver:
if role client:
netperf:
send_size: 10K
test: SCTP_STREAM_MANY
queue: bisect
testbox: lkp-hsw-d01
tbox_group: lkp-hsw-d01
kconfig: x86_64-rhel
enqueue_time: 2016-05-13 04:37:43.657077933 +08:00
compiler: gcc-4.9
rootfs: debian-x86_64-2015-02-07.cgz
id: c19f48e6aabc57bbdcae935982a46b1135b06798
user: lkp
head_commit: 75032e56e9d83095df75b6029a764280f081a258
base_commit: 44549e8f5eea4e0a41b487b63e616cb089922b99
branch: linux-devel/devel-hourly-2016051213
node_roles: server client
result_root:
"/result/netperf/performance-ipv4-300s-200%-cs-localhost-10K-SCTP_STREAM_MANY/lkp-hsw-d01/debian-x86_64-2015-02-07.cgz/x86_64-rhel/gcc-4.9/2159197d66770ec01f75c93fb11dc66df81fd45b/0"
job_file:
"/lkp/scheduled/lkp-hsw-d01/bisect_netperf-performance-ipv4-300s-200%-cs-localhost-10K-SCTP_STREAM_MANY-debian-x86_64-2015-02-07.cgz-x86_64-rhel-2159197d66770ec01f75c93fb11dc66df81fd45b-20160513-73416-1q3uv8m-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-hsw-d01/bisect_netperf-performance-ipv4-300s-200%-cs-localhost-10K-SCTP_STREAM_MANY-debian-x86_64-2015-02-07.cgz-x86_64-rhel-2159197d66770ec01f75c93fb11dc66df81fd45b-20160513-73416-1q3uv8m-0.yaml
- ARCH=x86_64
- kconfig=x86_64-rhel
- branch=linux-devel/devel-hourly-2016051213
- commit=2159197d66770ec01f75c93fb11dc66df81fd45b
-
BOOT_IMAGE=/pkg/linux/x86_64-rhel/gcc-4.9/2159197d66770ec01f75c93fb11dc66df81fd45b/vmlinuz-4.6.0-rc5-00034-g2159197
- max_uptime=1500
-
RESULT_ROOT=/result/netperf/performance-ipv4-300s-200%-cs-localhost-10K-SCTP_STREAM_MANY/lkp-hsw-d01/debian-x86_64-2015-02-07.cgz/x86_64-rhel/gcc-4.9/2159197d66770ec01f75c93fb11dc66df81fd45b/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-4.9/2159197d66770ec01f75c93fb11dc66df81fd45b/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,/lkp/benchmarks/perf-profile-x86_64.cgz,/osimage/deps/debian-x86_64-2015-02-07.cgz/netperf.cgz,/lkp/benchmarks/netperf.cgz"
linux_headers_initrd:
"/pkg/linux/x86_64-rhel/gcc-4.9/2159197d66770ec01f75c93fb11dc66df81fd45b/linux-headers.cgz"
repeat_to: 2
kernel:
"/pkg/linux/x86_64-rhel/gcc-4.9/2159197d66770ec01f75c93fb11dc66df81fd45b/vmlinuz-4.6.0-rc5-00034-g2159197"
dequeue_time: 2016-05-13 04:48:10.882705107 +08:00
job_state: finished
loadavg: 18.84 12.78 5.63 1/157 4597
start_time: '1463086128'
end_time: '1463086429'
version: "/lkp/lkp/.src-20160513-001235"
2016-05-13 04:48:47 echo performance >
/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor
2016-05-13 04:48:47 echo performance >
/sys/devices/system/cpu/cpu1/cpufreq/scaling_governor
2016-05-13 04:48:47 echo performance >
/sys/devices/system/cpu/cpu2/cpufreq/scaling_governor
2016-05-13 04:48:47 echo performance >
/sys/devices/system/cpu/cpu3/cpufreq/scaling_governor
2016-05-13 04:48:47 echo performance >
/sys/devices/system/cpu/cpu4/cpufreq/scaling_governor
2016-05-13 04:48:47 echo performance >
/sys/devices/system/cpu/cpu5/cpufreq/scaling_governor
2016-05-13 04:48:47 echo performance >
/sys/devices/system/cpu/cpu6/cpufreq/scaling_governor
2016-05-13 04:48:47 echo performance >
/sys/devices/system/cpu/cpu7/cpufreq/scaling_governor
2016-05-13 04:48:47 netserver -4
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
2016-05-13 04:48:49 netperf -4 -t SCTP_STREAM_MANY -c -C -l 300 -- -m 10K -H
127.0.0.1 &
