* [tip:sched/core] [sched/fair] e837456fdc: stress-ng.clock.ops_per_sec 3.4% regression
@ 2025-11-24 13:36 kernel test robot
0 siblings, 0 replies; only message in thread
From: kernel test robot @ 2025-11-24 13:36 UTC (permalink / raw)
To: Mel Gorman
Cc: oe-lkp, lkp, linux-kernel, x86, Peter Zijlstra, aubrey.li,
yu.c.chen, oliver.sang
Hello,
we reported an improvement in pts tests (also included in this report). now we
notice a small regression in stress-ng tests. just FYI.
kernel test robot noticed a 3.4% regression of stress-ng.clock.ops_per_sec on:
commit: e837456fdca81899a3c8e47b3fd39e30eae6e291 ("sched/fair: Reimplement NEXT_BUDDY to align with EEVDF goals")
https://git.kernel.org/cgit/linux/kernel/git/tip/tip.git sched/core
testcase: stress-ng
config: x86_64-rhel-9.4
compiler: gcc-14
test machine: 256 threads 2 sockets Intel(R) Xeon(R) 6768P CPU @ 2.4GHz (Granite Rapids) with 64G memory
parameters:
nr_threads: 100%
testtime: 60s
test: clock
cpufreq_governor: performance
In addition to that, the commit also has significant impact on the following tests:
+------------------+-----------------------------------------------------------------------------------------------+
| testcase: change | pts: pts.quadray.1.1080p.fps 23.2% improvement |
| test machine | 96 threads 2 sockets Intel(R) Xeon(R) Gold 6252 CPU @ 2.10GHz (Cascade Lake) with 512G memory |
| test parameters | cpufreq_governor=performance |
| | need_x=true |
| | option_a=5 |
| | option_b=1080p |
| | test=quadray-1.0.0 |
+------------------+-----------------------------------------------------------------------------------------------+
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@intel.com>
| Closes: https://lore.kernel.org/oe-lkp/202511242155.286a5a75-lkp@intel.com
Details are as below:
-------------------------------------------------------------------------------------------------->
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20251124/202511242155.286a5a75-lkp@intel.com
=========================================================================================
compiler/cpufreq_governor/kconfig/nr_threads/rootfs/tbox_group/test/testcase/testtime:
gcc-14/performance/x86_64-rhel-9.4/100%/debian-13-x86_64-20250902.cgz/lkp-gnr-2sp4/clock/stress-ng/60s
commit:
aceccac58a ("sched/fair: Enable scheduler feature NEXT_BUDDY")
e837456fdc ("sched/fair: Reimplement NEXT_BUDDY to align with EEVDF goals")
aceccac58ad76305 e837456fdca81899a3c8e47b3fd
---------------- ---------------------------
%stddev %change %stddev
\ | \
1.032e+09 ± 3% +28.0% 1.321e+09 ± 2% cpuidle..time
897841 ± 5% -15.0% 763285 ± 7% numa-numastat.node0.local_node
896471 ± 5% -14.8% 763770 ± 7% numa-vmstat.node0.numa_local
45743 +1.1% 46225 proc-vmstat.nr_kernel_stack
5.60 +2.0 7.57 mpstat.cpu.all.idle%
0.77 ± 3% +0.1 0.90 mpstat.cpu.all.irq%
59739 ± 2% -11.1% 53102 ± 4% perf-c2c.HITM.local
17302 ± 2% -9.8% 15614 ± 3% perf-c2c.HITM.remote
77041 ± 2% -10.8% 68716 ± 3% perf-c2c.HITM.total
5452857 -3.4% 5269019 stress-ng.clock.ops
90934 -3.4% 87863 stress-ng.clock.ops_per_sec
84662 +20.6% 102102 ± 3% stress-ng.time.involuntary_context_switches
24199 -2.3% 23648 stress-ng.time.percent_of_cpu_this_job_got
14530 -2.3% 14199 stress-ng.time.system_time
0.30 ± 44% +0.1 0.40 turbostat.C1%
0.95 ± 44% +0.4 1.34 ± 3% turbostat.C1E%
2.01 ± 44% +2.2 4.17 ± 3% turbostat.C6%
0.36 ± 44% +121.2% 0.80 ± 4% turbostat.CPU%c1
2.07 ± 48% +62.7% 3.37 ± 2% turbostat.CPU%c6
75763 -10.4% 67848 sched_debug.cfs_rq:/.avg_vruntime.avg
33867 ± 4% -15.8% 28530 ± 7% sched_debug.cfs_rq:/.avg_vruntime.min
75459 -10.4% 67579 sched_debug.cfs_rq:/.zero_vruntime.avg
33783 ± 4% -16.3% 28291 ± 7% sched_debug.cfs_rq:/.zero_vruntime.min
110597 ± 16% +41.9% 156958 ± 12% sched_debug.cpu.max_idle_balance_cost.stddev
-36.17 -30.4% -25.17 sched_debug.cpu.nr_uninterruptible.min
14.46 ± 6% -20.0% 11.57 ± 10% sched_debug.cpu.nr_uninterruptible.stddev
4.239e+10 -2.3% 4.14e+10 perf-stat.i.branch-instructions
0.08 +0.0 0.08 perf-stat.i.branch-miss-rate%
33247700 +2.5% 34067265 perf-stat.i.branch-misses
28.18 -0.8 27.42 perf-stat.i.cache-miss-rate%
90091009 -3.5% 86913480 perf-stat.i.cache-misses
3.205e+08 -1.0% 3.175e+08 perf-stat.i.cache-references
8.768e+11 -2.2% 8.572e+11 perf-stat.i.cpu-cycles
1621 -8.3% 1486 perf-stat.i.cpu-migrations
9765 +1.6% 9920 perf-stat.i.cycles-between-cache-misses
2.121e+11 -2.4% 2.07e+11 perf-stat.i.instructions
0.42 -1.2% 0.42 perf-stat.overall.MPKI
0.08 +0.0 0.08 perf-stat.overall.branch-miss-rate%
28.11 -0.7 27.37 perf-stat.overall.cache-miss-rate%
9732 +1.4% 9867 perf-stat.overall.cycles-between-cache-misses
4.168e+10 -2.3% 4.07e+10 perf-stat.ps.branch-instructions
32683973 +2.4% 33483421 perf-stat.ps.branch-misses
88576640 -3.6% 85423812 perf-stat.ps.cache-misses
3.151e+08 -0.9% 3.121e+08 perf-stat.ps.cache-references
8.62e+11 -2.2% 8.429e+11 perf-stat.ps.cpu-cycles
1594 -8.2% 1463 perf-stat.ps.cpu-migrations
2.085e+11 -2.4% 2.036e+11 perf-stat.ps.instructions
1.262e+13 -2.3% 1.233e+13 perf-stat.total.instructions
***************************************************************************************************
lkp-csl-2sp7: 96 threads 2 sockets Intel(R) Xeon(R) Gold 6252 CPU @ 2.10GHz (Cascade Lake) with 512G memory
=========================================================================================
compiler/cpufreq_governor/kconfig/need_x/option_a/option_b/rootfs/tbox_group/test/testcase:
gcc-14/performance/x86_64-rhel-9.4/true/5/1080p/debian-12-x86_64-phoronix/lkp-csl-2sp7/quadray-1.0.0/pts
commit:
aceccac58a ("sched/fair: Enable scheduler feature NEXT_BUDDY")
e837456fdc ("sched/fair: Reimplement NEXT_BUDDY to align with EEVDF goals")
aceccac58ad76305 e837456fdca81899a3c8e47b3fd
---------------- ---------------------------
%stddev %change %stddev
\ | \
15179434 ± 18% +51.8% 23046485 ± 15% meminfo.DirectMap2M
0.37 +0.0 0.41 mpstat.cpu.all.sys%
1055135 ± 31% -53.4% 492047 ± 91% numa-meminfo.node0.Shmem
263798 ± 31% -53.4% 122995 ± 91% numa-vmstat.node0.nr_shmem
21104 +9.2% 23039 vmstat.system.cs
371194 +1.8% 378021 proc-vmstat.nr_shmem
4628 +2.9% 4762 proc-vmstat.numa_huge_pte_updates
18161 ± 6% +22.2% 22195 ± 11% sched_debug.cfs_rq:/system.slice.avg_vruntime.stddev
18051 ± 6% +22.4% 22090 ± 11% sched_debug.cfs_rq:/system.slice.zero_vruntime.stddev
16231 +11.2% 18053 sched_debug.cpu.nr_switches.avg
6.26 ± 10% -19.7% 5.03 ± 10% sched_debug.cpu.nr_uninterruptible.stddev
39.69 ± 2% +23.2% 48.89 pts.quadray.1.1080p.fps
3245 +2.4% 3323 pts.time.percent_of_cpu_this_job_got
39.97 ± 2% +14.1% 45.62 pts.time.system_time
4456 +2.2% 4555 pts.time.user_time
1290701 +9.9% 1419060 pts.time.voluntary_context_switches
6.13e+09 +2.8% 6.299e+09 perf-stat.i.branch-instructions
21517 +9.0% 23449 perf-stat.i.context-switches
6.949e+10 +2.3% 7.108e+10 perf-stat.i.cpu-cycles
203.07 +7.8% 218.95 perf-stat.i.cpu-migrations
8980 ± 2% -7.8% 8279 perf-stat.i.cycles-between-cache-misses
2.515e+10 +2.9% 2.589e+10 perf-stat.i.dTLB-loads
8.21e+09 +2.7% 8.429e+09 perf-stat.i.dTLB-stores
6.904e+10 +2.9% 7.102e+10 perf-stat.i.instructions
0.72 +2.3% 0.74 perf-stat.i.metric.GHz
514.91 ± 2% -9.4% 466.42 ± 5% perf-stat.i.metric.K/sec
412.78 +2.9% 424.58 perf-stat.i.metric.M/sec
372305 ± 2% +20.8% 449603 ± 5% perf-stat.i.node-store-misses
1.81 -0.0 1.77 perf-stat.overall.branch-miss-rate%
57.23 ± 2% +4.3 61.51 ± 3% perf-stat.overall.node-store-miss-rate%
6.085e+09 +2.8% 6.254e+09 perf-stat.ps.branch-instructions
21348 +9.0% 23273 perf-stat.ps.context-switches
6.9e+10 +2.3% 7.059e+10 perf-stat.ps.cpu-cycles
201.61 +7.8% 217.38 perf-stat.ps.cpu-migrations
2.498e+10 +2.9% 2.571e+10 perf-stat.ps.dTLB-loads
8.154e+09 +2.7% 8.373e+09 perf-stat.ps.dTLB-stores
6.855e+10 +2.9% 7.053e+10 perf-stat.ps.instructions
369608 ± 2% +20.8% 446351 ± 5% perf-stat.ps.node-store-misses
9.519e+12 +3.0% 9.803e+12 perf-stat.total.instructions
0.00 ±223% +1490.9% 0.03 ±109% perf-sched.sch_delay.avg.ms.__cond_resched.smpboot_thread_fn.kthread.ret_from_fork.ret_from_fork_asm
0.01 ± 9% -94.7% 0.00 ±111% perf-sched.sch_delay.avg.ms.__cond_resched.ww_mutex_lock.drm_gem_vunmap.drm_gem_fb_vunmap.drm_atomic_helper_commit_planes
0.06 ± 21% -86.5% 0.01 ± 17% perf-sched.sch_delay.avg.ms.do_nanosleep.hrtimer_nanosleep.__x64_sys_nanosleep.do_syscall_64
0.12 ± 31% -89.2% 0.01 ± 68% perf-sched.sch_delay.avg.ms.do_task_dead.do_exit.do_group_exit.__x64_sys_exit_group.x64_sys_call
0.06 ± 67% -67.6% 0.02 ± 40% perf-sched.sch_delay.avg.ms.exit_to_user_mode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
0.03 ± 2% -36.3% 0.02 ± 3% perf-sched.sch_delay.avg.ms.futex_do_wait.__futex_wait.futex_wait.do_futex
0.21 ± 12% -87.8% 0.03 ± 7% perf-sched.sch_delay.avg.ms.irqentry_exit_to_user_mode.asm_sysvec_apic_timer_interrupt.[unknown].[unknown]
0.05 ± 14% -78.5% 0.01 ± 14% perf-sched.sch_delay.avg.ms.schedule_hrtimeout_range_clock.ep_poll.do_epoll_wait.do_epoll_pwait.part
0.03 ± 43% -70.0% 0.01 ± 9% perf-sched.sch_delay.avg.ms.schedule_hrtimeout_range_clock.poll_schedule_timeout.constprop.0.do_poll
0.22 ± 56% -94.6% 0.01 ± 41% perf-sched.sch_delay.avg.ms.schedule_hrtimeout_range_clock.poll_schedule_timeout.constprop.0.do_select
0.02 ± 27% -54.8% 0.01 ± 6% perf-sched.sch_delay.avg.ms.schedule_timeout.rcu_gp_fqs_loop.rcu_gp_kthread.kthread
0.00 ±223% +3636.4% 0.07 ±169% perf-sched.sch_delay.max.ms.__cond_resched.smpboot_thread_fn.kthread.ret_from_fork.ret_from_fork_asm
1.11 ± 92% -98.3% 0.02 ± 11% perf-sched.sch_delay.max.ms.__x64_sys_pause.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
2.06 ± 6% -98.4% 0.03 ± 15% perf-sched.sch_delay.max.ms.do_nanosleep.hrtimer_nanosleep.__x64_sys_nanosleep.do_syscall_64
3.22 ± 4% -63.6% 1.17 ± 98% perf-sched.sch_delay.max.ms.do_task_dead.do_exit.do_group_exit.__x64_sys_exit_group.x64_sys_call
0.67 ±129% +808.5% 6.12 ± 52% perf-sched.sch_delay.max.ms.exit_to_user_mode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe.[unknown]
0.00 ±223% +4066.7% 0.04 ±153% perf-sched.sch_delay.max.ms.irqentry_exit_to_user_mode.asm_exc_page_fault.[unknown]
0.02 ± 32% +715.6% 0.12 ±149% perf-sched.sch_delay.max.ms.irqentry_exit_to_user_mode.asm_sysvec_call_function_single.[unknown]
2.22 ± 14% -43.5% 1.25 ± 30% perf-sched.sch_delay.max.ms.schedule_hrtimeout_range_clock.ep_poll.do_epoll_wait.do_epoll_pwait.part
1.84 ± 48% -91.5% 0.16 ±169% perf-sched.sch_delay.max.ms.schedule_hrtimeout_range_clock.poll_schedule_timeout.constprop.0.do_poll
1.68 ± 55% -94.8% 0.09 ±164% perf-sched.sch_delay.max.ms.schedule_hrtimeout_range_clock.poll_schedule_timeout.constprop.0.do_select
0.35 ±130% -93.5% 0.02 ± 25% perf-sched.sch_delay.max.ms.schedule_timeout.kcompactd.kthread.ret_from_fork
1.34 ± 35% -97.2% 0.04 ± 48% perf-sched.sch_delay.max.ms.schedule_timeout.rcu_gp_fqs_loop.rcu_gp_kthread.kthread
0.03 ± 3% -37.6% 0.02 ± 3% perf-sched.total_sch_delay.average.ms
8.33 ± 4% -18.9% 6.76 ± 3% perf-sched.total_wait_and_delay.average.ms
162856 +20.3% 195924 perf-sched.total_wait_and_delay.count.ms
8.30 ± 4% -18.8% 6.74 ± 3% perf-sched.total_wait_time.average.ms
213.84 ± 51% -87.7% 26.25 ± 77% perf-sched.wait_and_delay.avg.ms.__cond_resched.generic_perform_write.shmem_file_write_iter.vfs_write.ksys_write
15.84 ± 2% -100.0% 0.00 perf-sched.wait_and_delay.avg.ms.__cond_resched.ww_mutex_lock.drm_gem_vunmap.drm_gem_fb_vunmap.drm_atomic_helper_commit_planes
800.53 -15.3% 677.82 ± 13% perf-sched.wait_and_delay.avg.ms.do_nanosleep.hrtimer_nanosleep.common_nsleep.__x64_sys_clock_nanosleep
2.37 ± 2% -21.9% 1.85 perf-sched.wait_and_delay.avg.ms.futex_do_wait.__futex_wait.futex_wait.do_futex
1.38 ± 9% -51.8% 0.67 ± 42% perf-sched.wait_and_delay.avg.ms.irqentry_exit_to_user_mode.asm_sysvec_apic_timer_interrupt.[unknown].[unknown]
38.24 ± 7% -12.2% 33.58 ± 7% perf-sched.wait_and_delay.avg.ms.schedule_hrtimeout_range_clock.ep_poll.do_epoll_wait.__x64_sys_epoll_wait
61.60 ± 3% -24.2% 46.70 ± 2% perf-sched.wait_and_delay.avg.ms.schedule_hrtimeout_range_clock.poll_schedule_timeout.constprop.0.do_poll
13.33 ± 24% +217.5% 42.33 ± 9% perf-sched.wait_and_delay.count.__cond_resched.generic_perform_write.shmem_file_write_iter.vfs_write.ksys_write
0.33 ±141% +5300.0% 18.00 ± 12% perf-sched.wait_and_delay.count.__cond_resched.mutex_lock.perf_poll.do_poll.constprop
196.50 ± 2% -100.0% 0.00 perf-sched.wait_and_delay.count.__cond_resched.ww_mutex_lock.drm_gem_vunmap.drm_gem_fb_vunmap.drm_atomic_helper_commit_planes
152066 ± 2% +19.5% 181763 perf-sched.wait_and_delay.count.futex_do_wait.__futex_wait.futex_wait.do_futex
1290 ± 6% +27.7% 1647 ± 12% perf-sched.wait_and_delay.count.irqentry_exit_to_user_mode.asm_sysvec_apic_timer_interrupt.[unknown].[unknown]
272.83 +16.9% 319.00 ± 2% perf-sched.wait_and_delay.count.schedule_hrtimeout_range_clock.ep_poll.do_epoll_wait.__x64_sys_epoll_wait
238.33 ± 4% +24.8% 297.33 perf-sched.wait_and_delay.count.schedule_hrtimeout_range_clock.poll_schedule_timeout.constprop.0.do_poll
2356 ± 47% -66.5% 788.64 ± 92% perf-sched.wait_and_delay.max.ms.__cond_resched.generic_perform_write.shmem_file_write_iter.vfs_write.ksys_write
44.95 ±147% +2131.0% 1002 ± 69% perf-sched.wait_and_delay.max.ms.__cond_resched.mutex_lock.perf_poll.do_poll.constprop
59.34 ± 41% -100.0% 0.00 perf-sched.wait_and_delay.max.ms.__cond_resched.ww_mutex_lock.drm_gem_vunmap.drm_gem_fb_vunmap.drm_atomic_helper_commit_planes
0.01 ±146% +30486.1% 1.84 ± 31% perf-sched.wait_time.avg.ms.__cond_resched.dput.__fput.__x64_sys_close.do_syscall_64
213.84 ± 51% -87.7% 26.25 ± 77% perf-sched.wait_time.avg.ms.__cond_resched.generic_perform_write.shmem_file_write_iter.vfs_write.ksys_write
15.82 ± 2% -96.0% 0.63 ± 66% perf-sched.wait_time.avg.ms.__cond_resched.ww_mutex_lock.drm_gem_vunmap.drm_gem_fb_vunmap.drm_atomic_helper_commit_planes
800.27 -15.3% 677.81 ± 13% perf-sched.wait_time.avg.ms.do_nanosleep.hrtimer_nanosleep.common_nsleep.__x64_sys_clock_nanosleep
0.61 ± 5% +15.0% 0.70 ± 7% perf-sched.wait_time.avg.ms.do_wait.kernel_wait4.do_syscall_64.entry_SYSCALL_64_after_hwframe
2.34 ± 2% -21.8% 1.83 perf-sched.wait_time.avg.ms.futex_do_wait.__futex_wait.futex_wait.do_futex
1.05 ± 2% +199.9% 3.15 perf-sched.wait_time.avg.ms.irqentry_exit_to_user_mode.asm_sysvec_apic_timer_interrupt.[unknown]
0.11 ± 17% +1066.9% 1.23 ± 12% perf-sched.wait_time.avg.ms.irqentry_exit_to_user_mode.asm_sysvec_call_function_single.[unknown]
0.47 ± 38% +148.3% 1.17 ± 30% perf-sched.wait_time.avg.ms.irqentry_exit_to_user_mode.asm_sysvec_reschedule_ipi.[unknown]
38.23 ± 7% -12.2% 33.58 ± 7% perf-sched.wait_time.avg.ms.schedule_hrtimeout_range_clock.ep_poll.do_epoll_wait.__x64_sys_epoll_wait
61.57 ± 3% -24.2% 46.69 ± 2% perf-sched.wait_time.avg.ms.schedule_hrtimeout_range_clock.poll_schedule_timeout.constprop.0.do_poll
0.00 ± 49% +5108.3% 0.10 ± 73% perf-sched.wait_time.avg.ms.wait_for_partner.fifo_open.do_dentry_open.vfs_open
0.01 ±142% +28030.6% 2.30 ± 23% perf-sched.wait_time.max.ms.__cond_resched.dput.__fput.__x64_sys_close.do_syscall_64
2356 ± 47% -66.5% 788.64 ± 92% perf-sched.wait_time.max.ms.__cond_resched.generic_perform_write.shmem_file_write_iter.vfs_write.ksys_write
45.07 ±146% +2125.3% 1002 ± 69% perf-sched.wait_time.max.ms.__cond_resched.mutex_lock.perf_poll.do_poll.constprop
11.69 ± 6% -17.3% 9.67 ± 9% perf-sched.wait_time.max.ms.do_nanosleep.hrtimer_nanosleep.__x64_sys_nanosleep.do_syscall_64
237.73 ±143% +561.4% 1572 ± 61% perf-sched.wait_time.max.ms.futex_do_wait.__futex_wait.futex_wait.do_futex
3.69 ± 4% +1614.4% 63.33 ± 48% perf-sched.wait_time.max.ms.irqentry_exit_to_user_mode.asm_sysvec_call_function_single.[unknown]
3.90 ± 26% +959.8% 41.38 ±100% perf-sched.wait_time.max.ms.irqentry_exit_to_user_mode.asm_sysvec_call_function_single.[unknown].[unknown]
0.04 ± 50% +5947.9% 2.19 ± 52% perf-sched.wait_time.max.ms.wait_for_partner.fifo_open.do_dentry_open.vfs_open
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:[~2025-11-24 13:37 UTC | newest]
Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-11-24 13:36 [tip:sched/core] [sched/fair] e837456fdc: stress-ng.clock.ops_per_sec 3.4% regression 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.