| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <202511251755.6e00cfb9-lkp@intel.com>
Date: Tue, 25 Nov 2025 17:41:00 +0800
From: kerne test robot <oliver.sang@...el.com>
To: Peter Zijlstra <peterz@...radead.org>
CC: <oe-lkp@...ts.linux.dev>, <lkp@...el.com>, <linux-kernel@...r.kernel.org>,
<x86@...nel.org>, Dietmar Eggemann <dietmar.eggemann@....com>, Chris Mason
<clm@...a.com>, <aubrey.li@...ux.intel.com>, <yu.c.chen@...el.com>,
<oliver.sang@...el.com>
Subject: [tip:sched/core] [sched/fair] 33cf66d883: vm-scalability.throughput
3.9% improvement
Hello,
kernel test robot noticed a 3.9% improvement of vm-scalability.throughput on:
commit: 33cf66d88306663d16e4759e9d24766b0aaa2e17 ("sched/fair: Proportional newidle balance")
https://git.kernel.org/cgit/linux/kernel/git/tip/tip.git sched/core
testcase: vm-scalability
config: x86_64-rhel-9.4
compiler: gcc-14
test machine: 224 threads 4 sockets Intel(R) Xeon(R) Platinum 8380H CPU @ 2.90GHz (Cooper Lake) with 192G memory
parameters:
runtime: 300s
size: 8T
test: anon-w-seq-mt
cpufreq_governor: performance
In addition to that, the commit also has significant impact on the following tests:
+------------------+----------------------------------------------------------------------------------------------------+
| testcase: change | vm-scalability: vm-scalability.throughput 6.9% improvement |
| test machine | 224 threads 4 sockets Intel(R) Xeon(R) Platinum 8380H CPU @ 2.90GHz (Cooper Lake) with 192G memory |
| test parameters | cpufreq_governor=performance |
| | runtime=300s |
| | size=8T |
| | test=anon-w-seq |
+------------------+----------------------------------------------------------------------------------------------------+
Details are as below:
-------------------------------------------------------------------------------------------------->
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20251125/202511251755.6e00cfb9-lkp@intel.com
=========================================================================================
compiler/cpufreq_governor/kconfig/rootfs/runtime/size/tbox_group/test/testcase:
gcc-14/performance/x86_64-rhel-9.4/debian-13-x86_64-20250902.cgz/300s/8T/lkp-cpl-4sp2/anon-w-seq-mt/vm-scalability
commit:
08d473dd87 ("sched/fair: Small cleanup to update_newidle_cost()")
33cf66d883 ("sched/fair: Proportional newidle balance")
08d473dd8718e4a4 33cf66d88306663d16e4759e9d2
---------------- ---------------------------
%stddev %change %stddev
\ | \
140169 ± 11% +20.4% 168807 ± 11% numa-meminfo.node0.Slab
38.84 +3.1 41.95 turbostat.C1%
35883 ± 16% -38.7% 22013 ± 12% sched_debug.cpu.max_idle_balance_cost.stddev
0.25 ± 33% -55.9% 0.11 ± 58% sched_debug.cpu.nr_uninterruptible.avg
-89.12 -21.9% -69.57 sched_debug.cpu.nr_uninterruptible.min
32696319 +205.2% 99805183 sched_debug.sysctl_sched.sysctl_sched_features
30.68 -0.4 30.28 perf-profile.calltrace.cycles-pp.do_rw_once
0.59 -0.0 0.56 perf-profile.calltrace.cycles-pp.asm_sysvec_apic_timer_interrupt.do_rw_once
0.07 ± 8% -0.0 0.02 ± 99% perf-profile.children.cycles-pp.schedule
0.08 ± 6% -0.0 0.05 perf-profile.children.cycles-pp.sched_balance_rq
0.08 ± 8% -0.0 0.06 ± 6% perf-profile.children.cycles-pp.__schedule
0.06 -0.0 0.05 perf-profile.self.cycles-pp.___perf_sw_event
0.05 +0.0 0.06 perf-profile.self.cycles-pp.mas_walk
0.08 ± 2% -10.1% 0.07 vm-scalability.free_time
348488 +1.7% 354569 vm-scalability.median
93790815 +3.9% 97486053 vm-scalability.throughput
383366 ± 5% +27.3% 487878 ± 2% vm-scalability.time.involuntary_context_switches
13116 -5.1% 12451 vm-scalability.time.percent_of_cpu_this_job_got
18386 -4.5% 17565 vm-scalability.time.system_time
21178 -5.2% 20069 vm-scalability.time.user_time
274218 -2.3% 267786 vm-scalability.time.voluntary_context_switches
8334443 -6.2% 7821786 proc-vmstat.nr_active_anon
8182472 -6.2% 7672252 proc-vmstat.nr_anon_pages
15772 -6.3% 14776 proc-vmstat.nr_anon_transparent_hugepages
3714681 +1.1% 3753946 proc-vmstat.nr_dirty_background_threshold
7438445 +1.1% 7517071 proc-vmstat.nr_dirty_threshold
37387239 +1.0% 37779422 proc-vmstat.nr_free_pages
37148190 +1.1% 37546282 proc-vmstat.nr_free_pages_blocks
17965 -5.9% 16898 proc-vmstat.nr_page_table_pages
8334431 -6.2% 7821774 proc-vmstat.nr_zone_active_anon
3.38 -2.0% 3.31 perf-stat.i.MPKI
6.728e+10 -2.0% 6.591e+10 perf-stat.i.branch-instructions
0.06 -0.0 0.06 ± 2% perf-stat.i.branch-miss-rate%
26644187 -11.1% 23698025 ± 2% perf-stat.i.branch-misses
66.58 -1.2 65.43 perf-stat.i.cache-miss-rate%
7.144e+08 -3.9% 6.868e+08 perf-stat.i.cache-misses
1.07e+09 -2.3% 1.045e+09 perf-stat.i.cache-references
8866 ± 2% +3.7% 9196 perf-stat.i.context-switches
2.51 -4.4% 2.40 perf-stat.i.cpi
5.33e+11 -6.2% 4.999e+11 perf-stat.i.cpu-cycles
665.04 -8.8% 606.57 perf-stat.i.cpu-migrations
743.32 -2.5% 724.57 perf-stat.i.cycles-between-cache-misses
2.109e+11 -2.1% 2.066e+11 perf-stat.i.instructions
0.40 +4.7% 0.42 perf-stat.i.ipc
3.39 -2.0% 3.33 perf-stat.overall.MPKI
0.03 -0.0 0.03 perf-stat.overall.branch-miss-rate%
66.92 -1.2 65.75 perf-stat.overall.cache-miss-rate%
2.53 -4.6% 2.42 perf-stat.overall.cpi
746.31 -2.6% 726.93 perf-stat.overall.cycles-between-cache-misses
0.39 +4.8% 0.41 perf-stat.overall.ipc
22267708 -8.1% 20466034 perf-stat.ps.branch-misses
6.891e+08 -2.4% 6.726e+08 perf-stat.ps.cache-misses
8639 +3.3% 8927 perf-stat.ps.context-switches
5.143e+11 -4.9% 4.889e+11 perf-stat.ps.cpu-cycles
624.80 -6.6% 583.36 perf-stat.ps.cpu-migrations
***************************************************************************************************
lkp-cpl-4sp2: 224 threads 4 sockets Intel(R) Xeon(R) Platinum 8380H CPU @ 2.90GHz (Cooper Lake) with 192G memory
=========================================================================================
compiler/cpufreq_governor/kconfig/rootfs/runtime/size/tbox_group/test/testcase:
gcc-14/performance/x86_64-rhel-9.4/debian-13-x86_64-20250902.cgz/300s/8T/lkp-cpl-4sp2/anon-w-seq/vm-scalability
commit:
08d473dd87 ("sched/fair: Small cleanup to update_newidle_cost()")
33cf66d883 ("sched/fair: Proportional newidle balance")
08d473dd8718e4a4 33cf66d88306663d16e4759e9d2
---------------- ---------------------------
%stddev %change %stddev
\ | \
12868898 ± 8% -13.6% 11117210 ± 6% meminfo.DirectMap2M
22.51 ± 2% +5.1 27.56 mpstat.cpu.all.idle%
4570 ± 6% -8.2% 4195 ± 5% perf-c2c.DRAM.remote
32696319 +205.2% 99805183 sched_debug.sysctl_sched.sysctl_sched_features
25712 +13.8% 29272 uptime.idle
1.544e+10 ± 2% +22.7% 1.895e+10 cpuidle..time
16705139 ± 2% +21.5% 20304654 cpuidle..usage
9686016 ± 9% -11.2% 8601707 ± 10% numa-meminfo.node0.AnonHugePages
9810366 ± 9% -11.5% 8680271 ± 10% numa-meminfo.node0.AnonPages
2.08 ± 10% +38.9% 2.88 ± 14% perf-sched.sch_delay.avg.ms.[unknown].[unknown].[unknown].[unknown].[unknown]
2.08 ± 10% +38.9% 2.88 ± 14% perf-sched.total_sch_delay.average.ms
22.68 ± 2% +5.2 27.82 turbostat.C1%
22.38 ± 2% +22.6% 27.43 turbostat.CPU%c1
5618917 -8.7% 5131336 ± 3% turbostat.NMI
847.70 -1.8% 832.15 turbostat.PkgWatt
43.20 -1.3% 42.66 turbostat.RAMWatt
0.01 -7.0% 0.01 vm-scalability.free_time
347285 +3.1% 358120 vm-scalability.median
2.87 ± 5% +0.5 3.42 ± 9% vm-scalability.median_stddev%
7.35 ± 6% +1.6 8.98 ± 6% vm-scalability.stddev%
82530869 +6.9% 88238114 vm-scalability.throughput
766040 ± 2% +6.1% 812920 vm-scalability.time.involuntary_context_switches
16652 -6.7% 15540 vm-scalability.time.percent_of_cpu_this_job_got
23121 -6.3% 21670 vm-scalability.time.system_time
27174 -7.0% 25266 vm-scalability.time.user_time
108375 +3.4% 112073 vm-scalability.time.voluntary_context_switches
9573443 -7.9% 8820226 proc-vmstat.nr_active_anon
9381538 -7.9% 8641925 proc-vmstat.nr_anon_pages
18112 -8.0% 16669 proc-vmstat.nr_anon_transparent_hugepages
3615325 +1.6% 3673671 proc-vmstat.nr_dirty_background_threshold
7239491 +1.6% 7356326 proc-vmstat.nr_dirty_threshold
36406902 +1.6% 36983177 proc-vmstat.nr_free_pages
36161545 +1.6% 36744053 proc-vmstat.nr_free_pages_blocks
18649 -5.6% 17609 proc-vmstat.nr_page_table_pages
145727 -4.0% 139928 ± 2% proc-vmstat.nr_shmem
9573428 -7.9% 8820214 proc-vmstat.nr_zone_active_anon
4676636 -12.0% 4113946 proc-vmstat.numa_huge_pte_updates
2.395e+09 -12.0% 2.106e+09 proc-vmstat.numa_pte_updates
3.69 -2.6% 3.59 perf-stat.i.MPKI
25933220 ± 2% -7.5% 23983956 perf-stat.i.branch-misses
72.64 -1.7 70.92 perf-stat.i.cache-miss-rate%
8.254e+08 -3.3% 7.985e+08 perf-stat.i.cache-misses
8791 +6.3% 9344 perf-stat.i.context-switches
3.00 -6.3% 2.81 perf-stat.i.cpi
6.733e+11 -7.0% 6.262e+11 perf-stat.i.cpu-cycles
812.82 -3.9% 781.10 perf-stat.i.cycles-between-cache-misses
0.34 +7.0% 0.36 perf-stat.i.ipc
3.70 -2.5% 3.61 perf-stat.overall.MPKI
0.03 -0.0 0.03 perf-stat.overall.branch-miss-rate%
72.83 -1.7 71.16 perf-stat.overall.cache-miss-rate%
3.01 -6.2% 2.83 perf-stat.overall.cpi
814.12 -3.8% 783.11 perf-stat.overall.cycles-between-cache-misses
0.33 +6.6% 0.35 perf-stat.overall.ipc
22052951 -7.8% 20340363 perf-stat.ps.branch-misses
7.949e+08 -2.8% 7.728e+08 perf-stat.ps.cache-misses
8696 +4.7% 9105 perf-stat.ps.context-switches
6.471e+11 -6.5% 6.052e+11 perf-stat.ps.cpu-cycles
569.36 +1.4% 577.21 perf-stat.ps.cpu-migrations
32.67 -0.9 31.80 perf-profile.calltrace.cycles-pp.do_rw_once
44.13 -0.4 43.74 perf-profile.calltrace.cycles-pp.asm_exc_page_fault.do_access
44.00 -0.4 43.62 perf-profile.calltrace.cycles-pp.exc_page_fault.asm_exc_page_fault.do_access
44.00 -0.4 43.62 perf-profile.calltrace.cycles-pp.do_user_addr_fault.exc_page_fault.asm_exc_page_fault.do_access
43.80 -0.4 43.41 perf-profile.calltrace.cycles-pp.__handle_mm_fault.handle_mm_fault.do_user_addr_fault.exc_page_fault.asm_exc_page_fault
43.86 -0.4 43.48 perf-profile.calltrace.cycles-pp.handle_mm_fault.do_user_addr_fault.exc_page_fault.asm_exc_page_fault.do_access
43.70 -0.4 43.32 perf-profile.calltrace.cycles-pp.__do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault.do_user_addr_fault.exc_page_fault
43.13 -0.4 42.77 perf-profile.calltrace.cycles-pp.vma_alloc_anon_folio_pmd.__do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault.do_user_addr_fault
39.61 -0.3 39.28 perf-profile.calltrace.cycles-pp.clear_page_erms.folio_zero_user.vma_alloc_anon_folio_pmd.__do_huge_pmd_anonymous_page.__handle_mm_fault
41.47 -0.3 41.16 perf-profile.calltrace.cycles-pp.folio_zero_user.vma_alloc_anon_folio_pmd.__do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault
1.33 ± 21% +1.1 2.48 ± 17% perf-profile.calltrace.cycles-pp.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle.cpu_startup_entry
1.34 ± 21% +1.1 2.48 ± 17% perf-profile.calltrace.cycles-pp.cpuidle_enter.cpuidle_idle_call.do_idle.cpu_startup_entry.start_secondary
1.33 ± 21% +1.1 2.48 ± 16% perf-profile.calltrace.cycles-pp.acpi_safe_halt.acpi_idle_do_entry.acpi_idle_enter.cpuidle_enter_state.cpuidle_enter
1.33 ± 21% +1.2 2.48 ± 17% perf-profile.calltrace.cycles-pp.acpi_idle_do_entry.acpi_idle_enter.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call
1.33 ± 21% +1.2 2.48 ± 17% perf-profile.calltrace.cycles-pp.acpi_idle_enter.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle
1.36 ± 21% +1.2 2.52 ± 16% perf-profile.calltrace.cycles-pp.cpu_startup_entry.start_secondary.common_startup_64
1.36 ± 21% +1.2 2.52 ± 16% perf-profile.calltrace.cycles-pp.do_idle.cpu_startup_entry.start_secondary.common_startup_64
1.36 ± 21% +1.2 2.52 ± 16% perf-profile.calltrace.cycles-pp.start_secondary.common_startup_64
1.35 ± 21% +1.2 2.51 ± 16% perf-profile.calltrace.cycles-pp.cpuidle_idle_call.do_idle.cpu_startup_entry.start_secondary.common_startup_64
1.37 ± 21% +1.2 2.53 ± 16% perf-profile.calltrace.cycles-pp.common_startup_64
2.50 ± 21% +2.2 4.71 ± 16% perf-profile.calltrace.cycles-pp.asm_sysvec_apic_timer_interrupt.pv_native_safe_halt.acpi_safe_halt.acpi_idle_do_entry.acpi_idle_enter
42.55 -0.5 42.01 perf-profile.children.cycles-pp.do_rw_once
44.04 -0.4 43.65 perf-profile.children.cycles-pp.exc_page_fault
44.19 -0.4 43.80 perf-profile.children.cycles-pp.asm_exc_page_fault
44.04 -0.4 43.65 perf-profile.children.cycles-pp.do_user_addr_fault
43.89 -0.4 43.51 perf-profile.children.cycles-pp.handle_mm_fault
43.82 -0.4 43.44 perf-profile.children.cycles-pp.__handle_mm_fault
43.70 -0.4 43.32 perf-profile.children.cycles-pp.__do_huge_pmd_anonymous_page
43.13 -0.4 42.77 perf-profile.children.cycles-pp.vma_alloc_anon_folio_pmd
42.23 -0.3 41.88 perf-profile.children.cycles-pp.folio_zero_user
39.95 -0.3 39.62 perf-profile.children.cycles-pp.clear_page_erms
0.08 -0.0 0.07 perf-profile.children.cycles-pp.___perf_sw_event
0.45 ± 12% +0.1 0.54 ± 12% perf-profile.children.cycles-pp.drm_atomic_helper_commit
0.45 ± 11% +0.1 0.54 ± 12% perf-profile.children.cycles-pp.drm_atomic_commit
0.48 ± 11% +0.1 0.57 ± 13% perf-profile.children.cycles-pp.worker_thread
0.45 ± 11% +0.1 0.54 ± 13% perf-profile.children.cycles-pp.drm_atomic_helper_dirtyfb
0.45 ± 11% +0.1 0.54 ± 13% perf-profile.children.cycles-pp.drm_fb_helper_damage_work
0.45 ± 11% +0.1 0.54 ± 13% perf-profile.children.cycles-pp.drm_fbdev_shmem_helper_fb_dirty
0.47 ± 11% +0.1 0.56 ± 12% perf-profile.children.cycles-pp.process_one_work
3.27 ± 9% +1.1 4.34 ± 9% perf-profile.children.cycles-pp.asm_sysvec_apic_timer_interrupt
1.34 ± 21% +1.2 2.49 ± 17% perf-profile.children.cycles-pp.acpi_safe_halt
1.34 ± 21% +1.2 2.49 ± 17% perf-profile.children.cycles-pp.pv_native_safe_halt
1.34 ± 21% +1.2 2.49 ± 17% perf-profile.children.cycles-pp.acpi_idle_do_entry
1.34 ± 21% +1.2 2.49 ± 17% perf-profile.children.cycles-pp.acpi_idle_enter
1.34 ± 21% +1.2 2.50 ± 16% perf-profile.children.cycles-pp.cpuidle_enter
1.34 ± 21% +1.2 2.50 ± 16% perf-profile.children.cycles-pp.cpuidle_enter_state
1.36 ± 21% +1.2 2.52 ± 16% perf-profile.children.cycles-pp.start_secondary
1.36 ± 21% +1.2 2.52 ± 16% perf-profile.children.cycles-pp.cpuidle_idle_call
1.37 ± 21% +1.2 2.53 ± 16% perf-profile.children.cycles-pp.common_startup_64
1.37 ± 21% +1.2 2.53 ± 16% perf-profile.children.cycles-pp.cpu_startup_entry
1.37 ± 21% +1.2 2.53 ± 16% perf-profile.children.cycles-pp.do_idle
40.98 -0.6 40.37 perf-profile.self.cycles-pp.do_rw_once
39.02 -0.3 38.71 perf-profile.self.cycles-pp.clear_page_erms
0.99 -0.0 0.94 perf-profile.self.cycles-pp.folio_zero_user
0.44 ± 12% +0.1 0.53 ± 12% perf-profile.self.cycles-pp.memcpy_toio
1.24 ± 21% +1.1 2.34 ± 16% perf-profile.self.cycles-pp.pv_native_safe_halt
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
Powered by blists - more mailing lists