* [linux-next:master] [udp] 6471658dc6: netperf.Throughput_Mbps 200.0% improvement
@ 2025-09-26 8:40 kernel test robot
0 siblings, 0 replies; only message in thread
From: kernel test robot @ 2025-09-26 8:40 UTC (permalink / raw)
To: Eric Dumazet
Cc: oe-lkp, lkp, Paolo Abeni, Willem de Bruijn, David Ahern,
Kuniyuki Iwashima, Jakub Kicinski, netdev, oliver.sang
Hello,
kernel test robot noticed a 200.0% improvement of netperf.Throughput_Mbps on:
commit: 6471658dc66c670580a7616e75f51b52917e7883 ("udp: use skb_attempt_defer_free()")
https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master
testcase: netperf
config: x86_64-rhel-9.4
compiler: gcc-14
test machine: 192 threads 2 sockets Intel(R) Xeon(R) 6740E CPU @ 2.4GHz (Sierra Forest) with 256G memory
parameters:
ip: ipv4
runtime: 300s
nr_threads: 50%
cluster: cs-localhost
test: UDP_STREAM
cpufreq_governor: performance
Details are as below:
-------------------------------------------------------------------------------------------------->
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20250926/202509261609.dec14b91-lkp@intel.com
=========================================================================================
cluster/compiler/cpufreq_governor/ip/kconfig/nr_threads/rootfs/runtime/tbox_group/test/testcase:
cs-localhost/gcc-14/performance/ipv4/x86_64-rhel-9.4/50%/debian-13-x86_64-20250902.cgz/300s/lkp-srf-2sp3/UDP_STREAM/netperf
commit:
3cd04c8f4a ("udp: make busylock per socket")
6471658dc6 ("udp: use skb_attempt_defer_free()")
3cd04c8f4afed71a 6471658dc66c670580a7616e75f
---------------- ---------------------------
%stddev %change %stddev
\ | \
6.079e+09 ± 4% +47.8% 8.983e+09 cpuidle..time
4.012e+08 +320.9% 1.689e+09 ± 2% cpuidle..usage
9360404 +22.3% 11449805 ± 4% numa-meminfo.node1.Active
9360396 +22.3% 11449799 ± 4% numa-meminfo.node1.Active(anon)
8894257 ± 3% +22.2% 10867440 ± 3% numa-meminfo.node1.Shmem
1.044e+09 ± 3% +206.8% 3.203e+09 numa-numastat.node0.local_node
1.044e+09 ± 3% +206.8% 3.204e+09 numa-numastat.node0.numa_hit
1.013e+09 ± 2% +218.0% 3.221e+09 numa-numastat.node1.local_node
1.013e+09 ± 2% +218.0% 3.221e+09 numa-numastat.node1.numa_hit
9.93 ± 5% +4.4 14.28 mpstat.cpu.all.idle%
0.59 +1.3 1.89 mpstat.cpu.all.irq%
1.77 +12.4 14.15 ± 2% mpstat.cpu.all.soft%
86.78 -18.8 67.94 mpstat.cpu.all.sys%
0.93 +0.8 1.74 mpstat.cpu.all.usr%
1.044e+09 ± 3% +206.9% 3.204e+09 numa-vmstat.node0.numa_hit
1.044e+09 ± 3% +206.9% 3.203e+09 numa-vmstat.node0.numa_local
2339319 +22.4% 2863299 ± 4% numa-vmstat.node1.nr_active_anon
2222754 ± 3% +22.3% 2717697 ± 3% numa-vmstat.node1.nr_shmem
2339318 +22.4% 2863297 ± 4% numa-vmstat.node1.nr_zone_active_anon
1.013e+09 ± 2% +218.1% 3.221e+09 numa-vmstat.node1.numa_hit
1.013e+09 ± 2% +218.1% 3.221e+09 numa-vmstat.node1.numa_local
9763805 ± 3% +21.6% 11869079 ± 3% meminfo.Active
9763788 ± 3% +21.6% 11869062 ± 3% meminfo.Active(anon)
805138 +15.5% 929863 meminfo.AnonPages
12584871 ± 2% +15.7% 14565534 ± 2% meminfo.Cached
9930753 ± 3% +21.2% 12038335 ± 3% meminfo.Committed_AS
16167194 +14.9% 18577687 ± 2% meminfo.Memused
8962742 ± 3% +22.1% 10943457 ± 3% meminfo.Shmem
16392623 +14.4% 18753189 ± 2% meminfo.max_used_kB
38913 +200.8% 117050 netperf.ThroughputBoth_Mbps
3735655 +200.8% 11236826 netperf.ThroughputBoth_total_Mbps
18515 +201.7% 55862 netperf.ThroughputRecv_Mbps
1777441 +201.7% 5362763 netperf.ThroughputRecv_total_Mbps
20398 +200.0% 61188 netperf.Throughput_Mbps
1958214 +200.0% 5874063 netperf.Throughput_total_Mbps
88004812 ± 8% -78.3% 19110782 ± 18% netperf.time.involuntary_context_switches
41333 +18.4% 48917 netperf.time.minor_page_faults
9067 -24.1% 6883 netperf.time.percent_of_cpu_this_job_got
27208 -25.1% 20391 netperf.time.system_time
158.64 +115.0% 341.02 netperf.time.user_time
2.139e+09 +200.8% 6.433e+09 netperf.workload
2441370 ± 3% +21.6% 2967589 ± 3% proc-vmstat.nr_active_anon
201274 +15.5% 232430 proc-vmstat.nr_anon_pages
6049392 -1.0% 5989283 proc-vmstat.nr_dirty_background_threshold
12113577 -1.0% 11993210 proc-vmstat.nr_dirty_threshold
3146655 ± 2% +15.7% 3641743 ± 2% proc-vmstat.nr_file_pages
60865222 -1.0% 60263229 proc-vmstat.nr_free_pages
60728673 -0.9% 60156480 proc-vmstat.nr_free_pages_blocks
2241122 ± 3% +22.1% 2736223 ± 3% proc-vmstat.nr_shmem
43088 +2.6% 44210 proc-vmstat.nr_slab_reclaimable
2441370 ± 3% +21.6% 2967589 ± 3% proc-vmstat.nr_zone_active_anon
46931 ± 26% +1165.5% 593929 ± 21% proc-vmstat.numa_hint_faults
35701 ± 34% +1506.1% 573389 ± 22% proc-vmstat.numa_hint_faults_local
2.057e+09 +212.3% 6.425e+09 proc-vmstat.numa_hit
2.057e+09 +212.3% 6.424e+09 proc-vmstat.numa_local
10954 ± 3% +77.0% 19391 ± 2% proc-vmstat.numa_pages_migrated
95835 ± 35% +588.2% 659561 ± 20% proc-vmstat.numa_pte_updates
1.641e+10 +212.8% 5.132e+10 proc-vmstat.pgalloc_normal
1186751 +45.6% 1727586 ± 7% proc-vmstat.pgfault
1.641e+10 +212.8% 5.132e+10 proc-vmstat.pgfree
10954 ± 3% +77.0% 19391 ± 2% proc-vmstat.pgmigrate_success
48468 +7.4% 52040 proc-vmstat.pgreuse
1.689e+10 +108.1% 3.514e+10 perf-stat.i.branch-instructions
38251661 +111.9% 81050915 perf-stat.i.branch-misses
0.92 ± 45% +3.6 4.57 ± 41% perf-stat.i.cache-miss-rate%
5.1e+09 -84.4% 7.944e+08 ± 13% perf-stat.i.cache-references
3325973 ± 2% +251.7% 11696843 ± 2% perf-stat.i.context-switches
6.70 -56.4% 2.92 perf-stat.i.cpi
5.565e+11 -3.1% 5.393e+11 perf-stat.i.cpu-cycles
2315 ± 14% +1249.5% 31253 ± 9% perf-stat.i.cpu-migrations
8.352e+10 +121.5% 1.85e+11 perf-stat.i.instructions
0.15 +124.3% 0.34 perf-stat.i.ipc
17.32 ± 2% +251.7% 60.92 ± 2% perf-stat.i.metric.K/sec
3553 +50.1% 5334 ± 8% perf-stat.i.minor-faults
3553 +50.1% 5334 ± 8% perf-stat.i.page-faults
0.62 ± 68% +3.8 4.40 ± 41% perf-stat.overall.cache-miss-rate%
6.66 -56.2% 2.92 perf-stat.overall.cpi
0.15 +128.5% 0.34 perf-stat.overall.ipc
11795 -26.6% 8659 perf-stat.overall.path-length
1.683e+10 +108.1% 3.502e+10 perf-stat.ps.branch-instructions
38128826 +111.9% 80783216 perf-stat.ps.branch-misses
5.083e+09 -84.4% 7.918e+08 ± 13% perf-stat.ps.cache-references
3315367 ± 2% +251.7% 11658619 ± 2% perf-stat.ps.context-switches
5.547e+11 -3.1% 5.375e+11 perf-stat.ps.cpu-cycles
2310 ± 14% +1247.8% 31146 ± 9% perf-stat.ps.cpu-migrations
8.326e+10 +121.4% 1.844e+11 perf-stat.ps.instructions
3534 +50.0% 5302 ± 8% perf-stat.ps.minor-faults
3534 +50.0% 5302 ± 8% perf-stat.ps.page-faults
2.523e+13 +120.8% 5.57e+13 perf-stat.total.instructions
26243720 -28.6% 18749089 sched_debug.cfs_rq:/.avg_vruntime.avg
28003592 -25.9% 20752934 ± 4% sched_debug.cfs_rq:/.avg_vruntime.max
25141866 -32.7% 16920920 sched_debug.cfs_rq:/.avg_vruntime.min
0.30 ± 4% +26.3% 0.38 ± 2% sched_debug.cfs_rq:/.h_nr_queued.stddev
0.30 ± 4% +28.8% 0.38 ± 2% sched_debug.cfs_rq:/.h_nr_runnable.stddev
209424 ± 34% +169.4% 564215 ± 12% sched_debug.cfs_rq:/.left_deadline.avg
2071516 ± 26% +50.9% 3126519 ± 6% sched_debug.cfs_rq:/.left_deadline.stddev
209420 ± 34% +169.4% 564200 ± 12% sched_debug.cfs_rq:/.left_vruntime.avg
2071477 ± 26% +50.9% 3126440 ± 6% sched_debug.cfs_rq:/.left_vruntime.stddev
26243720 -28.6% 18749089 sched_debug.cfs_rq:/.min_vruntime.avg
28003592 -25.9% 20752934 ± 4% sched_debug.cfs_rq:/.min_vruntime.max
25141866 -32.7% 16920920 sched_debug.cfs_rq:/.min_vruntime.min
0.27 ± 3% +24.1% 0.33 ± 2% sched_debug.cfs_rq:/.nr_queued.stddev
209420 ± 34% +169.4% 564200 ± 12% sched_debug.cfs_rq:/.right_vruntime.avg
2071477 ± 26% +50.9% 3126440 ± 6% sched_debug.cfs_rq:/.right_vruntime.stddev
209.56 ± 6% +32.0% 276.63 ± 2% sched_debug.cfs_rq:/.runnable_avg.stddev
192.89 ± 5% +31.7% 253.94 ± 2% sched_debug.cfs_rq:/.util_avg.stddev
819031 ± 3% -65.5% 282924 ± 8% sched_debug.cpu.avg_idle.avg
4756 ± 2% -37.0% 2995 ± 2% sched_debug.cpu.avg_idle.min
1084177 -55.0% 487973 ± 11% sched_debug.cpu.avg_idle.stddev
740.98 ± 24% +55.0% 1148 ± 24% sched_debug.cpu.clock_task.stddev
2338 ± 5% +22.8% 2872 sched_debug.cpu.curr->pid.stddev
742501 ± 19% +46.3% 1085919 ± 5% sched_debug.cpu.max_idle_balance_cost.min
106154 ± 12% -44.8% 58635 ± 9% sched_debug.cpu.max_idle_balance_cost.stddev
0.00 ± 17% -20.6% 0.00 ± 3% sched_debug.cpu.next_balance.stddev
0.31 ± 5% +24.8% 0.39 sched_debug.cpu.nr_running.stddev
2579918 ± 2% +252.5% 9093241 ± 2% sched_debug.cpu.nr_switches.avg
3594985 +181.0% 10103415 ± 4% sched_debug.cpu.nr_switches.max
1042225 ± 52% +546.5% 6737845 ± 14% sched_debug.cpu.nr_switches.min
135.08 ± 23% -24.8% 101.56 ± 18% sched_debug.cpu.nr_uninterruptible.max
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.
--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2025-09-26 8:40 UTC | newest]
Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-09-26 8:40 [linux-next:master] [udp] 6471658dc6: netperf.Throughput_Mbps 200.0% improvement kernel test robot
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.