Re: [linus:master] [mm] 96204e1531: will-it-scale.per_thread_ops -3.9% regression

From: Ryan Roberts
Date: Thu Feb 08 2024 - 10:40:48 EST


On 08/02/2024 07:36, kernel test robot wrote:
>
>
> Hello,
>
> kernel test robot noticed a -3.9% regression of will-it-scale.per_thread_ops on:

Hi,

I've tried reproducing this on x86_64, but I see fluctuations of +/- 5% between
runs without changing the kernel, when running the page_fault1 test. So I'm not
sure a -3.9% regression is real?

The fingered change is actually a fix for a functional bug, so I'm not sure we
should be preferring performance over correctness anyway :)

I'm not planning to do anything further here, unless someone shouts that there
is a real problem.

>
>
> commit: 96204e15310c218fd9355bdcacd02fed1d18070e ("mm: thp_get_unmapped_area must honour topdown preference")
> https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
>
> testcase: will-it-scale
> test machine: 224 threads 2 sockets Intel(R) Xeon(R) Platinum 8480CTDX (Sapphire Rapids) with 512G memory
> parameters:
>
> test: page_fault1
> cpufreq_governor: performance
>
>
>
>
> If you fix the issue in a separate patch/commit (i.e. not just a new version of
> the same patch/commit), kindly add following tags
> | Reported-by: kernel test robot <oliver.sang@xxxxxxxxx>
> | Closes: https://lore.kernel.org/oe-lkp/202402081435.c3faa4b4-oliver.sang@xxxxxxxxx
>
>
> Details are as below:
> -------------------------------------------------------------------------------------------------->
>
>
> The kernel config and materials to reproduce are available at:
> https://download.01.org/0day-ci/archive/20240208/202402081435.c3faa4b4-oliver.sang@xxxxxxxxx
>
> =========================================================================================
> compiler/cpufreq_governor/kconfig/rootfs/tbox_group/test/testcase:
> gcc-12/performance/x86_64-rhel-8.3/debian-11.1-x86_64-20220510.cgz/lkp-spr-2sp4/page_fault1/will-it-scale
>
> commit:
> 4ef9ad19e1 ("mm: huge_memory: don't force huge page alignment on 32 bit")
> 96204e1531 ("mm: thp_get_unmapped_area must honour topdown preference")
>

I don't really understand the below table. Is a positive or negative change
better? There appear to be tests with much more fluctuation than the reported
page_fault1 test. Are these real?

Thanks,
Ryan

> 4ef9ad19e17676b9 96204e15310c218fd9355bdcacd
> ---------------- ---------------------------
> %stddev %change %stddev
> \ | \
> 32.93 -3.5 29.43 mpstat.cpu.all.sys%
> 77.89 � 2% -10.2% 69.98 vmstat.procs.r
> 4678 +14.0% 5332 vmstat.system.cs
> 251960 -3.4% 243306 vmstat.system.in
> 1049 -10.2% 943.00 turbostat.Avg_MHz
> 36.22 -3.7 32.54 turbostat.Busy%
> 29.10 +1.8 30.87 turbostat.C1E%
> 0.35 +15.7% 0.40 turbostat.CPU%c6
> 565.62 -2.3% 552.53 turbostat.PkgWatt
> 123847 -18.4% 101096 meminfo.Active
> 123657 -18.3% 101000 meminfo.Active(anon)
> 8127530 +10.3% 8966990 meminfo.AnonHugePages
> 8298705 +10.3% 9156780 meminfo.AnonPages
> 8417195 +10.1% 9267954 meminfo.Inactive
> 8416461 +10.1% 9267742 meminfo.Inactive(anon)
> 138121 -16.9% 114720 meminfo.Shmem
> 4852175 � 2% +9.2% 5296708 numa-meminfo.node0.AnonHugePages
> 4941270 � 2% +9.0% 5387476 numa-meminfo.node0.AnonPages
> 121571 -18.9% 98652 numa-meminfo.node1.Active
> 121528 -18.9% 98620 numa-meminfo.node1.Active(anon)
> 3270590 +12.4% 3674514 numa-meminfo.node1.AnonHugePages
> 3352686 +12.6% 3773824 numa-meminfo.node1.AnonPages
> 3395079 +12.4% 3815833 numa-meminfo.node1.Inactive
> 3394971 +12.4% 3815762 numa-meminfo.node1.Inactive(anon)
> 125041 -18.4% 102036 numa-meminfo.node1.Shmem
> 1219136 � 2% +10.3% 1344427 numa-vmstat.node0.nr_anon_pages
> 2337 +10.4% 2581 numa-vmstat.node0.nr_anon_transparent_hugepages
> 1237208 � 2% +10.1% 1362084 numa-vmstat.node0.nr_inactive_anon
> 1236994 � 2% +10.1% 1362125 numa-vmstat.node0.nr_zone_inactive_anon
> 30361 -18.8% 24656 numa-vmstat.node1.nr_active_anon
> 828422 +13.5% 940280 numa-vmstat.node1.nr_anon_pages
> 1577 +13.3% 1788 numa-vmstat.node1.nr_anon_transparent_hugepages
> 838873 +13.3% 950629 � 2% numa-vmstat.node1.nr_inactive_anon
> 31244 -18.4% 25510 numa-vmstat.node1.nr_shmem
> 30361 -18.8% 24656 numa-vmstat.node1.nr_zone_active_anon
> 838696 +13.4% 950674 � 2% numa-vmstat.node1.nr_zone_inactive_anon
> 30907 -18.4% 25233 proc-vmstat.nr_active_anon
> 2068862 +11.0% 2295458 proc-vmstat.nr_anon_pages
> 3957 +11.0% 4390 proc-vmstat.nr_anon_transparent_hugepages
> 2099316 +10.6% 2322543 proc-vmstat.nr_inactive_anon
> 7379 +4.4% 7701 proc-vmstat.nr_page_table_pages
> 34526 -17.0% 28665 proc-vmstat.nr_shmem
> 30907 -18.4% 25233 proc-vmstat.nr_zone_active_anon
> 2099307 +10.6% 2322537 proc-vmstat.nr_zone_inactive_anon
> 26039 � 10% -16.4% 21774 � 6% proc-vmstat.numa_hint_faults_local
> 51127959 -3.7% 49228917 proc-vmstat.numa_hit
> 50891640 -3.7% 48994322 proc-vmstat.numa_local
> 32315 -21.8% 25269 proc-vmstat.pgactivate
> 1.265e+10 -3.8% 1.217e+10 proc-vmstat.pgalloc_normal
> 25947640 -3.6% 25012103 proc-vmstat.pgfault
> 1.265e+10 -3.8% 1.217e+10 proc-vmstat.pgfree
> 24650837 -3.8% 23717158 proc-vmstat.thp_fault_alloc
> 44754589 -15.5% 37839290 will-it-scale.112.threads
> 74.20 +11.2% 82.49 will-it-scale.112.threads_idle
> 36349325 -6.0% 34182802 will-it-scale.168.threads
> 72.79 +11.7% 81.28 will-it-scale.168.threads_idle
> 40356275 -5.6% 38083547 will-it-scale.224.threads
> 66.97 +19.5% 80.06 will-it-scale.224.threads_idle
> 28146956 -15.5% 23782499 will-it-scale.56.threads
> 78.16 +8.4% 84.70 will-it-scale.56.threads_idle
> 795871 -3.9% 764972 will-it-scale.per_thread_ops
> 67103 -4.9% 63834 will-it-scale.time.involuntary_context_switches
> 17508243 � 4% +6.9% 18724480 will-it-scale.time.maximum_resident_set_size
> 22382078 -3.6% 21577203 will-it-scale.time.minor_page_faults
> 7050 -7.3% 6534 will-it-scale.time.percent_of_cpu_this_job_got
> 20998 -7.4% 19448 will-it-scale.time.system_time
> 1075 -3.3% 1040 will-it-scale.time.user_time
> 236326 +40.3% 331522 will-it-scale.time.voluntary_context_switches
> 4.056e+08 -3.9% 3.899e+08 will-it-scale.workload
> 5880710 -9.8% 5305863 sched_debug.cfs_rq:/.avg_vruntime.avg
> 10300738 � 2% -11.0% 9163982 � 2% sched_debug.cfs_rq:/.avg_vruntime.max
> 1772532 � 6% -15.5% 1498526 � 4% sched_debug.cfs_rq:/.avg_vruntime.min
> 2066050 -12.1% 1817081 sched_debug.cfs_rq:/.avg_vruntime.stddev
> 321.14 � 36% -63.9% 115.93 � 57% sched_debug.cfs_rq:/.load_avg.avg
> 1945 � 20% -47.8% 1015 � 50% sched_debug.cfs_rq:/.load_avg.stddev
> 5880710 -9.8% 5305863 sched_debug.cfs_rq:/.min_vruntime.avg
> 10300738 � 2% -11.0% 9163982 � 2% sched_debug.cfs_rq:/.min_vruntime.max
> 1772532 � 6% -15.5% 1498526 � 4% sched_debug.cfs_rq:/.min_vruntime.min
> 2066050 -12.1% 1817081 sched_debug.cfs_rq:/.min_vruntime.stddev
> 233.57 � 8% -24.3% 176.83 � 3% sched_debug.cfs_rq:/.runnable_avg.avg
> 958.36 � 4% -9.2% 870.11 � 3% sched_debug.cfs_rq:/.runnable_avg.max
> 240.06 � 5% -20.4% 190.99 � 5% sched_debug.cfs_rq:/.runnable_avg.stddev
> 232.74 � 9% -24.3% 176.20 � 3% sched_debug.cfs_rq:/.util_avg.avg
> 238.62 � 5% -20.4% 189.97 � 5% sched_debug.cfs_rq:/.util_avg.stddev
> 134.97 � 20% -46.6% 72.12 � 9% sched_debug.cfs_rq:/.util_est.avg
> 211.62 � 7% -30.9% 146.18 � 3% sched_debug.cfs_rq:/.util_est.stddev
> 85860 � 4% +20.2% 103165 � 4% sched_debug.cpu.avg_idle.stddev
> 2360 � 15% -32.2% 1600 � 6% sched_debug.cpu.curr->pid.avg
> 4127 � 5% -11.4% 3655 � 2% sched_debug.cpu.curr->pid.stddev
> 0.18 � 13% -30.8% 0.13 � 5% sched_debug.cpu.nr_running.avg
> 0.35 � 3% -11.4% 0.31 sched_debug.cpu.nr_running.stddev
> 4008 +10.9% 4445 sched_debug.cpu.nr_switches.avg
> 0.25 � 9% +22.4% 0.31 � 2% sched_debug.cpu.nr_uninterruptible.avg
> 234.42 +10.7% 259.59 perf-stat.i.MPKI
> 1.038e+09 -2.6% 1.012e+09 perf-stat.i.branch-instructions
> 0.45 +0.0 0.47 perf-stat.i.branch-miss-rate%
> 1.192e+09 +4.1% 1.241e+09 perf-stat.i.cache-misses
> 1.307e+09 +3.2% 1.349e+09 perf-stat.i.cache-references
> 4550 +15.1% 5238 perf-stat.i.context-switches
> 37.73 -11.3% 33.45 perf-stat.i.cpi
> 2.282e+11 -10.6% 2.041e+11 perf-stat.i.cpu-cycles
> 395.06 -4.0% 379.37 perf-stat.i.cycles-between-cache-misses
> 0.08 -0.0 0.07 perf-stat.i.dTLB-load-miss-rate%
> 957972 -11.5% 848218 perf-stat.i.dTLB-load-misses
> 1.142e+09 -2.6% 1.112e+09 perf-stat.i.dTLB-loads
> 4327055 -3.0% 4195482 perf-stat.i.dTLB-store-misses
> 9.982e+08 -3.2% 9.66e+08 perf-stat.i.dTLB-stores
> 5.057e+09 -2.7% 4.92e+09 perf-stat.i.instructions
> 0.08 +4.6% 0.09 perf-stat.i.ipc
> 1.02 -10.5% 0.91 perf-stat.i.metric.GHz
> 245.40 -1.7% 241.21 perf-stat.i.metric.K/sec
> 19.83 -1.0% 19.63 perf-stat.i.metric.M/sec
> 81125 -3.9% 77994 perf-stat.i.minor-faults
> 9961580 -4.6% 9501596 perf-stat.i.node-loads
> 81127 -3.9% 77996 perf-stat.i.page-faults
> 239.06 +6.3% 254.04 perf-stat.overall.MPKI
> 0.41 +0.0 0.42 perf-stat.overall.branch-miss-rate%
> 45.33 -7.8% 41.81 perf-stat.overall.cpi
> 189.64 -13.2% 164.59 perf-stat.overall.cycles-between-cache-misses
> 0.09 -0.0 0.08 perf-stat.overall.dTLB-load-miss-rate%
> 0.02 +8.4% 0.02 perf-stat.overall.ipc
> 3930 +1.4% 3987 perf-stat.overall.path-length
> 1.044e+09 -2.5% 1.017e+09 perf-stat.ps.branch-instructions
> 1.215e+09 +3.5% 1.257e+09 perf-stat.ps.cache-misses
> 1.329e+09 +2.7% 1.365e+09 perf-stat.ps.cache-references
> 4559 +14.8% 5231 perf-stat.ps.context-switches
> 2.304e+11 -10.2% 2.069e+11 perf-stat.ps.cpu-cycles
> 977167 -11.5% 864759 perf-stat.ps.dTLB-load-misses
> 1.146e+09 -2.5% 1.117e+09 perf-stat.ps.dTLB-loads
> 4393673 -3.1% 4258515 perf-stat.ps.dTLB-store-misses
> 1.006e+09 -3.2% 9.745e+08 perf-stat.ps.dTLB-stores
> 5.082e+09 -2.6% 4.948e+09 perf-stat.ps.instructions
> 82026 -3.8% 78940 perf-stat.ps.minor-faults
> 10075965 -4.4% 9629094 perf-stat.ps.node-loads
> 82028 -3.8% 78942 perf-stat.ps.page-faults
> 1.594e+12 -2.5% 1.555e+12 perf-stat.total.instructions
> 82.91 -2.8 80.13 perf-profile.calltrace.cycles-pp.clear_page_erms.clear_huge_page.__do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault
> 94.27 -1.2 93.10 perf-profile.calltrace.cycles-pp.testcase
> 92.74 -1.2 91.59 perf-profile.calltrace.cycles-pp.__handle_mm_fault.handle_mm_fault.do_user_addr_fault.exc_page_fault.asm_exc_page_fault
> 92.76 -1.1 91.62 perf-profile.calltrace.cycles-pp.handle_mm_fault.do_user_addr_fault.exc_page_fault.asm_exc_page_fault.testcase
> 92.82 -1.1 91.69 perf-profile.calltrace.cycles-pp.exc_page_fault.asm_exc_page_fault.testcase
> 92.82 -1.1 91.69 perf-profile.calltrace.cycles-pp.do_user_addr_fault.exc_page_fault.asm_exc_page_fault.testcase
> 94.24 -1.0 93.19 perf-profile.calltrace.cycles-pp.asm_exc_page_fault.testcase
> 90.44 -1.0 89.46 perf-profile.calltrace.cycles-pp.clear_huge_page.__do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault.do_user_addr_fault
> 90.76 -0.9 89.81 perf-profile.calltrace.cycles-pp.__do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault.do_user_addr_fault.exc_page_fault
> 0.59 � 2% -0.2 0.34 � 70% perf-profile.calltrace.cycles-pp.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.clear_page_erms.clear_huge_page.__do_huge_pmd_anonymous_page
> 1.98 -0.2 1.77 perf-profile.calltrace.cycles-pp.do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault.do_user_addr_fault.exc_page_fault
> 1.98 -0.2 1.76 perf-profile.calltrace.cycles-pp.vma_alloc_folio.do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault.do_user_addr_fault
> 1.80 -0.2 1.60 perf-profile.calltrace.cycles-pp.prep_compound_page.get_page_from_freelist.__alloc_pages.alloc_pages_mpol.vma_alloc_folio
> 1.85 -0.2 1.65 perf-profile.calltrace.cycles-pp.__alloc_pages.alloc_pages_mpol.vma_alloc_folio.do_huge_pmd_anonymous_page.__handle_mm_fault
> 1.84 -0.2 1.64 perf-profile.calltrace.cycles-pp.get_page_from_freelist.__alloc_pages.alloc_pages_mpol.vma_alloc_folio.do_huge_pmd_anonymous_page
> 1.85 -0.2 1.66 perf-profile.calltrace.cycles-pp.alloc_pages_mpol.vma_alloc_folio.do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault
> 0.98 -0.1 0.84 perf-profile.calltrace.cycles-pp.asm_sysvec_apic_timer_interrupt.clear_page_erms.clear_huge_page.__do_huge_pmd_anonymous_page.__handle_mm_fault
> 1.33 -0.1 1.28 perf-profile.calltrace.cycles-pp.free_unref_page_prepare.free_unref_page.release_pages.tlb_batch_pages_flush.tlb_finish_mmu
> 1.51 -0.0 1.46 perf-profile.calltrace.cycles-pp.tlb_finish_mmu.unmap_region.do_vmi_align_munmap.do_vmi_munmap.__vm_munmap
> 1.47 -0.0 1.43 perf-profile.calltrace.cycles-pp.tlb_batch_pages_flush.tlb_finish_mmu.unmap_region.do_vmi_align_munmap.do_vmi_munmap
> 1.38 -0.0 1.33 perf-profile.calltrace.cycles-pp.free_unref_page.release_pages.tlb_batch_pages_flush.tlb_finish_mmu.unmap_region
> 1.46 -0.0 1.42 perf-profile.calltrace.cycles-pp.release_pages.tlb_batch_pages_flush.tlb_finish_mmu.unmap_region.do_vmi_align_munmap
> 1.64 -0.0 1.60 perf-profile.calltrace.cycles-pp.unmap_region.do_vmi_align_munmap.do_vmi_munmap.__vm_munmap.__x64_sys_munmap
> 0.78 � 3% +0.2 0.96 perf-profile.calltrace.cycles-pp.intel_idle_xstate.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle
> 0.64 +0.2 0.82 � 3% perf-profile.calltrace.cycles-pp.__hrtimer_run_queues.hrtimer_interrupt.__sysvec_apic_timer_interrupt.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt
> 0.82 � 3% +0.3 1.12 � 4% perf-profile.calltrace.cycles-pp.hrtimer_interrupt.__sysvec_apic_timer_interrupt.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state
> 0.84 � 3% +0.3 1.16 � 4% perf-profile.calltrace.cycles-pp.__sysvec_apic_timer_interrupt.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state.cpuidle_enter
> 0.26 �100% +0.4 0.70 perf-profile.calltrace.cycles-pp.intel_idle.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle
> 1.26 � 5% +0.5 1.74 � 5% perf-profile.calltrace.cycles-pp.sysvec_apic_timer_interrupt.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call
> 1.40 � 4% +0.5 1.92 � 4% perf-profile.calltrace.cycles-pp.asm_sysvec_apic_timer_interrupt.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle
> 3.45 +0.9 4.35 perf-profile.calltrace.cycles-pp.__cond_resched.clear_huge_page.__do_huge_pmd_anonymous_page.__handle_mm_fault.handle_mm_fault
> 2.75 +1.0 3.70 � 2% perf-profile.calltrace.cycles-pp.cpuidle_enter_state.cpuidle_enter.cpuidle_idle_call.do_idle.cpu_startup_entry
> 2.88 +1.0 3.87 � 2% perf-profile.calltrace.cycles-pp.cpuidle_enter.cpuidle_idle_call.do_idle.cpu_startup_entry.start_secondary
> 3.08 +1.0 4.12 � 2% perf-profile.calltrace.cycles-pp.cpuidle_idle_call.do_idle.cpu_startup_entry.start_secondary.secondary_startup_64_no_verify
> 3.12 +1.1 4.18 � 2% perf-profile.calltrace.cycles-pp.do_idle.cpu_startup_entry.start_secondary.secondary_startup_64_no_verify
> 3.12 +1.1 4.18 � 2% perf-profile.calltrace.cycles-pp.cpu_startup_entry.start_secondary.secondary_startup_64_no_verify
> 3.12 +1.1 4.18 � 2% perf-profile.calltrace.cycles-pp.start_secondary.secondary_startup_64_no_verify
> 3.15 +1.1 4.22 � 2% perf-profile.calltrace.cycles-pp.secondary_startup_64_no_verify
> 84.16 -2.9 81.22 perf-profile.children.cycles-pp.clear_page_erms
> 92.75 -1.2 91.60 perf-profile.children.cycles-pp.__handle_mm_fault
> 92.77 -1.1 91.63 perf-profile.children.cycles-pp.handle_mm_fault
> 94.98 -1.1 93.84 perf-profile.children.cycles-pp.testcase
> 92.83 -1.1 91.70 perf-profile.children.cycles-pp.do_user_addr_fault
> 92.83 -1.1 91.70 perf-profile.children.cycles-pp.exc_page_fault
> 93.55 -1.1 92.47 perf-profile.children.cycles-pp.asm_exc_page_fault
> 90.13 -1.0 89.13 perf-profile.children.cycles-pp.clear_huge_page
> 90.76 -0.9 89.81 perf-profile.children.cycles-pp.__do_huge_pmd_anonymous_page
> 1.98 -0.2 1.77 perf-profile.children.cycles-pp.do_huge_pmd_anonymous_page
> 1.98 -0.2 1.76 perf-profile.children.cycles-pp.vma_alloc_folio
> 1.80 -0.2 1.60 perf-profile.children.cycles-pp.prep_compound_page
> 1.88 -0.2 1.69 perf-profile.children.cycles-pp.get_page_from_freelist
> 1.94 -0.2 1.76 perf-profile.children.cycles-pp.__alloc_pages
> 1.94 -0.2 1.76 perf-profile.children.cycles-pp.alloc_pages_mpol
> 0.38 -0.1 0.25 � 2% perf-profile.children.cycles-pp.asm_sysvec_call_function
> 1.38 -0.0 1.33 perf-profile.children.cycles-pp.free_unref_page_prepare
> 1.51 -0.0 1.46 perf-profile.children.cycles-pp.tlb_finish_mmu
> 1.39 -0.0 1.34 perf-profile.children.cycles-pp.free_unref_page
> 1.47 -0.0 1.42 perf-profile.children.cycles-pp.release_pages
> 1.47 -0.0 1.43 perf-profile.children.cycles-pp.tlb_batch_pages_flush
> 1.64 -0.0 1.60 perf-profile.children.cycles-pp.unmap_region
> 0.21 � 3% -0.0 0.18 � 3% perf-profile.children.cycles-pp.task_tick_fair
> 0.12 � 3% -0.0 0.10 � 4% perf-profile.children.cycles-pp.folio_prep_large_rmappable
> 0.09 � 4% +0.0 0.10 � 4% perf-profile.children.cycles-pp.free_tail_page_prepare
> 0.17 � 2% +0.0 0.19 � 3% perf-profile.children.cycles-pp.pte_alloc_one
> 0.17 � 4% +0.0 0.19 perf-profile.children.cycles-pp.perf_rotate_context
> 0.06 � 11% +0.0 0.08 � 8% perf-profile.children.cycles-pp._raw_spin_lock_irq
> 0.06 � 7% +0.0 0.08 � 5% perf-profile.children.cycles-pp.update_sg_lb_stats
> 0.06 � 8% +0.0 0.08 � 6% perf-profile.children.cycles-pp.clockevents_program_event
> 0.12 � 4% +0.0 0.14 � 3% perf-profile.children.cycles-pp.__intel_pmu_enable_all
> 0.16 � 4% +0.0 0.18 � 3% perf-profile.children.cycles-pp.native_irq_return_iret
> 0.08 � 4% +0.0 0.10 � 4% perf-profile.children.cycles-pp.find_busiest_group
> 0.08 � 6% +0.0 0.10 � 4% perf-profile.children.cycles-pp.update_sd_lb_stats
> 0.05 +0.0 0.08 � 8% perf-profile.children.cycles-pp.__schedule
> 0.11 � 3% +0.0 0.14 � 3% perf-profile.children.cycles-pp.load_balance
> 0.17 � 8% +0.0 0.21 � 3% perf-profile.children.cycles-pp.menu_select
> 0.01 �223% +0.0 0.06 � 6% perf-profile.children.cycles-pp.native_sched_clock
> 0.00 +0.1 0.05 perf-profile.children.cycles-pp.irqtime_account_irq
> 0.00 +0.1 0.05 perf-profile.children.cycles-pp.run_rebalance_domains
> 0.00 +0.1 0.05 perf-profile.children.cycles-pp.rwsem_optimistic_spin
> 0.00 +0.1 0.05 perf-profile.children.cycles-pp.update_blocked_averages
> 0.00 +0.1 0.05 � 7% perf-profile.children.cycles-pp.__mmap
> 0.00 +0.1 0.05 � 7% perf-profile.children.cycles-pp.vm_mmap_pgoff
> 0.00 +0.1 0.06 � 8% perf-profile.children.cycles-pp.lapic_next_deadline
> 0.11 � 4% +0.1 0.17 � 4% perf-profile.children.cycles-pp.rebalance_domains
> 0.00 +0.1 0.06 � 11% perf-profile.children.cycles-pp._raw_spin_trylock
> 0.00 +0.1 0.06 � 6% perf-profile.children.cycles-pp.down_write_killable
> 0.00 +0.1 0.06 � 6% perf-profile.children.cycles-pp.read_tsc
> 0.00 +0.1 0.06 � 6% perf-profile.children.cycles-pp.schedule
> 0.79 � 4% +0.1 0.87 � 3% perf-profile.children.cycles-pp.tick_nohz_highres_handler
> 0.19 � 3% +0.1 0.27 � 3% perf-profile.children.cycles-pp.__do_softirq
> 0.21 � 3% +0.1 0.31 � 3% perf-profile.children.cycles-pp.irq_exit_rcu
> 0.00 +0.1 0.10 � 3% perf-profile.children.cycles-pp.rwsem_down_write_slowpath
> 1.17 +0.1 1.29 perf-profile.children.cycles-pp.__hrtimer_run_queues
> 0.79 � 3% +0.2 0.97 perf-profile.children.cycles-pp.intel_idle_xstate
> 0.50 � 3% +0.2 0.70 perf-profile.children.cycles-pp.intel_idle
> 1.45 � 2% +0.2 1.67 � 3% perf-profile.children.cycles-pp.hrtimer_interrupt
> 1.48 � 2% +0.2 1.73 � 3% perf-profile.children.cycles-pp.__sysvec_apic_timer_interrupt
> 2.50 � 2% +0.4 2.89 � 3% perf-profile.children.cycles-pp.asm_sysvec_apic_timer_interrupt
> 1.94 � 4% +0.4 2.35 � 4% perf-profile.children.cycles-pp.sysvec_apic_timer_interrupt
> 3.60 +0.9 4.50 perf-profile.children.cycles-pp.__cond_resched
> 2.91 +1.0 3.90 � 2% perf-profile.children.cycles-pp.cpuidle_enter
> 2.90 +1.0 3.90 � 2% perf-profile.children.cycles-pp.cpuidle_enter_state
> 3.11 +1.0 4.15 � 2% perf-profile.children.cycles-pp.cpuidle_idle_call
> 3.12 +1.1 4.18 � 2% perf-profile.children.cycles-pp.start_secondary
> 3.15 +1.1 4.22 � 2% perf-profile.children.cycles-pp.cpu_startup_entry
> 3.15 +1.1 4.22 � 2% perf-profile.children.cycles-pp.do_idle
> 3.15 +1.1 4.22 � 2% perf-profile.children.cycles-pp.secondary_startup_64_no_verify
> 83.46 -2.9 80.61 perf-profile.self.cycles-pp.clear_page_erms
> 1.79 -0.2 1.59 perf-profile.self.cycles-pp.prep_compound_page
> 1.30 -0.1 1.25 perf-profile.self.cycles-pp.free_unref_page_prepare
> 0.12 � 3% -0.0 0.10 � 4% perf-profile.self.cycles-pp.folio_prep_large_rmappable
> 0.05 +0.0 0.06 � 7% perf-profile.self.cycles-pp.update_sg_lb_stats
> 0.06 � 11% +0.0 0.08 � 8% perf-profile.self.cycles-pp._raw_spin_lock_irq
> 0.12 � 4% +0.0 0.14 � 3% perf-profile.self.cycles-pp.__intel_pmu_enable_all
> 0.16 � 4% +0.0 0.18 � 3% perf-profile.self.cycles-pp.native_irq_return_iret
> 0.07 � 6% +0.0 0.10 � 5% perf-profile.self.cycles-pp.menu_select
> 0.20 � 3% +0.0 0.25 � 2% perf-profile.self.cycles-pp.cpuidle_enter_state
> 0.00 +0.1 0.06 � 8% perf-profile.self.cycles-pp.lapic_next_deadline
> 0.00 +0.1 0.06 � 6% perf-profile.self.cycles-pp.native_sched_clock
> 0.00 +0.1 0.06 � 13% perf-profile.self.cycles-pp._raw_spin_trylock
> 0.00 +0.1 0.06 � 9% perf-profile.self.cycles-pp.read_tsc
> 0.78 � 3% +0.2 0.97 perf-profile.self.cycles-pp.intel_idle_xstate
> 0.50 � 3% +0.2 0.70 perf-profile.self.cycles-pp.intel_idle
> 2.93 +0.9 3.83 perf-profile.self.cycles-pp.__cond_resched
> 2.59 +1.0 3.61 perf-profile.self.cycles-pp.clear_huge_page
>
>
>
>
> 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.
>
>