* [linus:master] [rseq] abc850e761: stress-ng.sem.sem_wait_calls_per_sec 3.1% improvement
@ 2025-12-09 15:41 kernel test robot
0 siblings, 0 replies; only message in thread
From: kernel test robot @ 2025-12-09 15:41 UTC (permalink / raw)
To: Thomas Gleixner
Cc: oe-lkp, lkp, linux-kernel, Ingo Molnar, Peter Zijlstra,
Mathieu Desnoyers, oliver.sang
Hello,
kernel test robot noticed a 3.1% improvement of stress-ng.sem.sem_wait_calls_per_sec on:
commit: abc850e7616c91ebaa3f5ba3617ab0a104d45039 ("rseq: Provide and use rseq_update_user_cs()")
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
testcase: stress-ng
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:
nr_threads: 100%
testtime: 60s
test: sem
cpufreq_governor: performance
Details are as below:
-------------------------------------------------------------------------------------------------->
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20251209/202512092342.3ee2de77-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-srf-2sp3/sem/stress-ng/60s
commit:
9c37cb6e80 ("rseq: Provide static branch for runtime debugging")
abc850e761 ("rseq: Provide and use rseq_update_user_cs()")
9c37cb6e80b8fcdd abc850e7616c91ebaa3f5ba3617
---------------- ---------------------------
%stddev %change %stddev
\ | \
713480 ± 29% -24.9% 536114 ± 28% meminfo.Mapped
19261235 ± 14% -28.3% 13815751 ± 45% perf-sched.total_wait_and_delay.count.ms
19261235 ± 14% -28.3% 13815751 ± 45% perf-sched.wait_and_delay.count.[unknown].[unknown].[unknown].[unknown].[unknown]
209285 ± 4% -3.8% 201417 ± 3% proc-vmstat.nr_anon_pages
179839 ± 29% -25.3% 134393 ± 28% proc-vmstat.nr_mapped
0.21 +0.0 0.22 perf-stat.i.branch-miss-rate%
3.044e+08 +3.6% 3.154e+08 perf-stat.i.branch-misses
1.933e+08 +3.5% 2.001e+08 perf-stat.i.context-switches
0.93 ± 3% +6.7% 0.99 perf-stat.i.metric.M/sec
0.20 +0.0 0.21 perf-stat.overall.branch-miss-rate%
2.996e+08 +3.6% 3.104e+08 perf-stat.ps.branch-misses
1.903e+08 +3.5% 1.97e+08 perf-stat.ps.context-switches
1.341e+10 +2.6% 1.377e+10 stress-ng.sem.ops
2.235e+08 +2.6% 2.294e+08 stress-ng.sem.ops_per_sec
374680 +3.1% 386364 stress-ng.sem.sem_timedwait_calls_per_sec
374638 +3.2% 386525 stress-ng.sem.sem_trywait_calls_per_sec
374649 +3.1% 386331 stress-ng.sem.sem_wait_calls_per_sec
1.178e+10 +3.5% 1.219e+10 stress-ng.time.involuntary_context_switches
7623 -1.2% 7530 stress-ng.time.system_time
3803 +2.8% 3908 stress-ng.time.user_time
8.44 -2.8 5.65 perf-profile.calltrace.cycles-pp.__rseq_handle_notify_resume.exit_to_user_mode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe.__sched_yield
9.62 -2.6 7.02 perf-profile.calltrace.cycles-pp.exit_to_user_mode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe.__sched_yield
59.80 -1.0 58.80 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.__sched_yield
60.34 -1.0 59.38 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.__sched_yield
0.83 +0.0 0.86 perf-profile.calltrace.cycles-pp._raw_spin_lock.raw_spin_rq_lock_nested.__schedule.schedule.__x64_sys_sched_yield
0.65 +0.0 0.68 perf-profile.calltrace.cycles-pp.__update_load_avg_se.update_load_avg.put_prev_entity.pick_next_task_fair.__pick_next_task
1.00 +0.0 1.03 perf-profile.calltrace.cycles-pp.update_load_avg.set_next_entity.pick_next_task_fair.__pick_next_task.__schedule
1.11 +0.0 1.15 perf-profile.calltrace.cycles-pp.__enqueue_entity.put_prev_entity.pick_next_task_fair.__pick_next_task.__schedule
1.56 +0.1 1.62 perf-profile.calltrace.cycles-pp.update_load_avg.put_prev_entity.pick_next_task_fair.__pick_next_task.__schedule
1.73 +0.1 1.79 perf-profile.calltrace.cycles-pp.native_sched_clock.sched_clock.sched_clock_cpu.update_rq_clock.yield_task_fair
1.77 +0.1 1.84 perf-profile.calltrace.cycles-pp.sched_clock.sched_clock_cpu.update_rq_clock.yield_task_fair.do_sched_yield
2.13 +0.1 2.20 perf-profile.calltrace.cycles-pp.__rdgsbase_inactive.__sched_yield
1.84 +0.1 1.90 perf-profile.calltrace.cycles-pp.sched_clock_cpu.update_rq_clock.yield_task_fair.do_sched_yield.__x64_sys_sched_yield
1.95 +0.1 2.02 perf-profile.calltrace.cycles-pp.set_next_entity.pick_next_task_fair.__pick_next_task.__schedule.schedule
2.06 +0.1 2.14 perf-profile.calltrace.cycles-pp.update_rq_clock.yield_task_fair.do_sched_yield.__x64_sys_sched_yield.do_syscall_64
2.38 +0.1 2.46 perf-profile.calltrace.cycles-pp.prepare_task_switch.__schedule.schedule.__x64_sys_sched_yield.do_syscall_64
2.98 +0.1 3.08 perf-profile.calltrace.cycles-pp.__wrgsbase_inactive.__sched_yield
3.26 +0.1 3.38 perf-profile.calltrace.cycles-pp.put_prev_entity.pick_next_task_fair.__pick_next_task.__schedule.schedule
5.15 +0.2 5.32 perf-profile.calltrace.cycles-pp.yield_task_fair.do_sched_yield.__x64_sys_sched_yield.do_syscall_64.entry_SYSCALL_64_after_hwframe
6.45 +0.2 6.66 perf-profile.calltrace.cycles-pp.do_sched_yield.__x64_sys_sched_yield.do_syscall_64.entry_SYSCALL_64_after_hwframe.__sched_yield
3.28 +0.2 3.50 perf-profile.calltrace.cycles-pp.rseq_update_cpu_node_id.__rseq_handle_notify_resume.exit_to_user_mode_loop.do_syscall_64.entry_SYSCALL_64_after_hwframe
9.21 +0.3 9.50 perf-profile.calltrace.cycles-pp.pick_next_task_fair.__pick_next_task.__schedule.schedule.__x64_sys_sched_yield
9.41 +0.3 9.70 perf-profile.calltrace.cycles-pp.__pick_next_task.__schedule.schedule.__x64_sys_sched_yield.do_syscall_64
17.41 +0.5 17.88 perf-profile.calltrace.cycles-pp.__schedule.schedule.__x64_sys_sched_yield.do_syscall_64.entry_SYSCALL_64_after_hwframe
17.59 +0.5 18.07 perf-profile.calltrace.cycles-pp.schedule.__x64_sys_sched_yield.do_syscall_64.entry_SYSCALL_64_after_hwframe.__sched_yield
16.02 +0.6 16.60 perf-profile.calltrace.cycles-pp.os_xsave.__sched_yield
24.16 +0.7 24.86 perf-profile.calltrace.cycles-pp.__x64_sys_sched_yield.do_syscall_64.entry_SYSCALL_64_after_hwframe.__sched_yield
20.98 +0.7 21.71 perf-profile.calltrace.cycles-pp.restore_fpregs_from_fpstate.switch_fpu_return.arch_exit_to_user_mode_prepare.do_syscall_64.entry_SYSCALL_64_after_hwframe
23.42 +0.8 24.26 perf-profile.calltrace.cycles-pp.switch_fpu_return.arch_exit_to_user_mode_prepare.do_syscall_64.entry_SYSCALL_64_after_hwframe.__sched_yield
25.22 +0.9 26.11 perf-profile.calltrace.cycles-pp.arch_exit_to_user_mode_prepare.do_syscall_64.entry_SYSCALL_64_after_hwframe.__sched_yield
8.52 -2.6 5.88 perf-profile.children.cycles-pp.__rseq_handle_notify_resume
9.64 -2.6 7.04 perf-profile.children.cycles-pp.exit_to_user_mode_loop
59.97 -1.0 58.99 perf-profile.children.cycles-pp.do_syscall_64
60.46 -1.0 59.50 perf-profile.children.cycles-pp.entry_SYSCALL_64_after_hwframe
0.09 +0.0 0.10 perf-profile.children.cycles-pp.propagate_entity_load_avg
0.12 ± 4% +0.0 0.14 ± 3% perf-profile.children.cycles-pp.clock_gettime
1.12 +0.0 1.16 perf-profile.children.cycles-pp.__enqueue_entity
1.14 +0.0 1.18 perf-profile.children.cycles-pp.__update_load_avg_se
1.97 +0.1 2.04 perf-profile.children.cycles-pp.set_next_entity
2.38 +0.1 2.46 perf-profile.children.cycles-pp.prepare_task_switch
2.41 +0.1 2.49 perf-profile.children.cycles-pp.__rdgsbase_inactive
2.67 +0.1 2.77 perf-profile.children.cycles-pp.update_load_avg
3.26 +0.1 3.37 perf-profile.children.cycles-pp.__wrgsbase_inactive
3.30 +0.1 3.41 perf-profile.children.cycles-pp.put_prev_entity
5.17 +0.2 5.34 perf-profile.children.cycles-pp.yield_task_fair
6.49 +0.2 6.70 perf-profile.children.cycles-pp.do_sched_yield
3.45 +0.2 3.68 perf-profile.children.cycles-pp.rseq_update_cpu_node_id
9.24 +0.3 9.54 perf-profile.children.cycles-pp.pick_next_task_fair
9.43 +0.3 9.73 perf-profile.children.cycles-pp.__pick_next_task
17.48 +0.5 17.96 perf-profile.children.cycles-pp.__schedule
17.61 +0.5 18.09 perf-profile.children.cycles-pp.schedule
16.04 +0.6 16.61 perf-profile.children.cycles-pp.os_xsave
24.18 +0.7 24.88 perf-profile.children.cycles-pp.__x64_sys_sched_yield
21.01 +0.7 21.75 perf-profile.children.cycles-pp.restore_fpregs_from_fpstate
23.44 +0.8 24.28 perf-profile.children.cycles-pp.switch_fpu_return
25.24 +0.9 26.15 perf-profile.children.cycles-pp.arch_exit_to_user_mode_prepare
0.67 +0.0 0.70 perf-profile.self.cycles-pp.___perf_sw_event
0.92 +0.0 0.95 perf-profile.self.cycles-pp.update_curr
1.11 +0.0 1.15 perf-profile.self.cycles-pp.__enqueue_entity
0.80 +0.0 0.84 perf-profile.self.cycles-pp.update_load_avg
0.70 +0.0 0.73 perf-profile.self.cycles-pp.pick_next_task_fair
1.04 +0.0 1.08 perf-profile.self.cycles-pp.exit_to_user_mode_loop
1.12 +0.1 1.17 perf-profile.self.cycles-pp.__update_load_avg_se
1.78 +0.1 1.84 perf-profile.self.cycles-pp.arch_exit_to_user_mode_prepare
1.48 +0.1 1.54 perf-profile.self.cycles-pp.prepare_task_switch
0.69 +0.1 0.75 perf-profile.self.cycles-pp.do_syscall_64
2.40 +0.1 2.48 perf-profile.self.cycles-pp.__rdgsbase_inactive
3.12 +0.1 3.23 perf-profile.self.cycles-pp.__wrgsbase_inactive
16.02 +0.6 16.61 perf-profile.self.cycles-pp.os_xsave
21.00 +0.7 21.74 perf-profile.self.cycles-pp.restore_fpregs_from_fpstate
0.37 +2.0 2.40 perf-profile.self.cycles-pp.__rseq_handle_notify_resume
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:[~2025-12-09 15:41 UTC | newest]
Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-12-09 15:41 [linus:master] [rseq] abc850e761: stress-ng.sem.sem_wait_calls_per_sec 3.1% 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.