Greeting,
FYI, we noticed a 4.1% improvement of aim9.link_test.ops_per_sec due to commit:
commit: 7ffddd499ba6122b1a07828f023d1d67629aa017 ("mm: hugetlb: fix a race between freeing and dissolving the page")
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
in testcase: aim9
on test machine: 256 threads Intel(R) Genuine Intel(R) CPU 0000 @ 1.30GHz with 112G memory
with following parameters:
testtime: 300s
test: link_test
cpufreq_governor: performance
ucode: 0xffff0190
test-description: Suite IX is the "AIM Independent Resource Benchmark:" the famous synthetic benchmark.
test-url: https://sourceforge.net/projects/aimbench/files/aim-suite9/
Details are as below:
-------------------------------------------------------------------------------------------------->
To reproduce:
git clone https://github.com/intel/lkp-tests.git
cd lkp-tests
bin/lkp install job.yaml # job file is attached in this email
bin/lkp split-job --compatible job.yaml
bin/lkp run compatible-job.yaml
=========================================================================================
compiler/cpufreq_governor/kconfig/rootfs/tbox_group/test/testcase/testtime/ucode:
gcc-9/performance/x86_64-rhel-8.3/debian-10.4-x86_64-20200603.cgz/lkp-knl-f1/link_test/aim9/300s/0xffff0190
commit:
585fc0d287 ("mm: hugetlbfs: fix cannot migrate the fallocated HugeTLB page")
7ffddd499b ("mm: hugetlb: fix a race between freeing and dissolving the page")
585fc0d2871c9318 7ffddd499ba6122b1a07828f023
---------------- ---------------------------
%stddev %change %stddev
\ | \
45758 +4.1% 47627 aim9.link_test.ops_per_sec
29210 ? 16% +85.5% 54200 ? 35% softirqs.CPU54.RCU
1323161 +1.7% 1345983 proc-vmstat.pgalloc_normal
1298032 +1.8% 1321221 proc-vmstat.pgfree
0.72 ?188% -96.2% 0.03 ?152% perf-sched.wait_time.avg.ms.schedule_timeout.wait_for_completion.stop_one_cpu.affine_move_task
0.29 ? 17% -41.2% 0.17 ? 26% perf-sched.wait_time.max.ms.preempt_schedule_common._cond_resched.shmem_undo_range.shmem_truncate_range.shmem_evict_inode
3.81 ?214% -98.7% 0.05 ?178% perf-sched.wait_time.max.ms.schedule_timeout.wait_for_completion.stop_one_cpu.affine_move_task
55.50 ? 19% +35.1% 75.00 ? 19% interrupts.CPU10.RES:Rescheduling_interrupts
223.33 +48.6% 331.83 ? 30% interrupts.CPU180.NMI:Non-maskable_interrupts
223.33 +48.6% 331.83 ? 30% interrupts.CPU180.PMI:Performance_monitoring_interrupts
210.17 ? 3% +194.4% 618.83 ? 60% interrupts.CPU191.NMI:Non-maskable_interrupts
210.17 ? 3% +194.4% 618.83 ? 60% interrupts.CPU191.PMI:Performance_monitoring_interrupts
189.00 ? 2% +48.3% 280.33 ? 31% interrupts.CPU247.NMI:Non-maskable_interrupts
189.00 ? 2% +48.3% 280.33 ? 31% interrupts.CPU247.PMI:Performance_monitoring_interrupts
291.33 ? 28% -34.5% 190.83 ? 2% interrupts.CPU38.NMI:Non-maskable_interrupts
291.33 ? 28% -34.5% 190.83 ? 2% interrupts.CPU38.PMI:Performance_monitoring_interrupts
58.33 ? 18% +55.1% 90.50 ? 13% interrupts.CPU52.RES:Rescheduling_interrupts
2.95 ? 3% -0.2 2.77 ? 3% perf-profile.calltrace.cycles-pp.tick_irq_enter.irq_enter_rcu.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state
1.39 ? 6% -0.1 1.25 ? 3% perf-profile.calltrace.cycles-pp.ktime_get.tick_irq_enter.irq_enter_rcu.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt
5.32 ? 2% -0.7 4.67 ? 4% perf-profile.children.cycles-pp.ktime_get
2.99 ? 3% -0.2 2.82 ? 3% perf-profile.children.cycles-pp.tick_irq_enter
0.53 ? 3% -0.1 0.41 ? 7% perf-profile.children.cycles-pp.hrtimer_forward
0.50 ? 4% -0.1 0.43 ? 5% perf-profile.children.cycles-pp.tick_check_broadcast_expired
0.08 ? 17% -0.1 0.03 ?102% perf-profile.children.cycles-pp.cpuidle_get_cpu_driver
0.42 ? 3% -0.0 0.40 ? 2% perf-profile.children.cycles-pp.rb_insert_color
0.21 ? 7% +0.0 0.26 ? 8% perf-profile.children.cycles-pp.balance_fair
0.45 ? 7% +0.1 0.52 ? 6% perf-profile.children.cycles-pp.schedule
4.07 ? 3% -0.6 3.46 ? 6% perf-profile.self.cycles-pp.ktime_get
0.52 ? 3% -0.1 0.41 ? 7% perf-profile.self.cycles-pp.hrtimer_forward
0.49 ? 4% -0.1 0.42 ? 5% perf-profile.self.cycles-pp.tick_check_broadcast_expired
0.41 ? 3% -0.0 0.38 ? 2% perf-profile.self.cycles-pp.rb_insert_color
0.09 ? 8% -0.0 0.07 ? 16% perf-profile.self.cycles-pp.cpuidle_reflect
0.42 ? 5% +0.0 0.46 ? 4% perf-profile.self.cycles-pp.get_next_timer_interrupt
aim9.link_test.ops_per_sec
51000 +-------------------------------------------------------------------+
| O |
50000 |-+ |
| O O |
| O O |
49000 |-+ O O O |
| O O O |
48000 |-+ O |
| O O O O O O OO O O O O |
47000 |-+ O O O O O O O O O |
| O O O O |
|.+.++.+.+. +. .+.++.+.+.+.++.+.+.+.++.+. ++. |
46000 |-+ +. : + +. + +. .++.+.+.+. |
| + + + + |
45000 +-------------------------------------------------------------------+
[*] 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,
Oliver Sang