Hello,
kernel test robot noticed a 3.9% improvement of will-it-scale.per_process_ops on:
commit: 5730609ffd7e558e1e3305d0c6839044e8f6591b ("select: do_pollfd: add unlikely branch hint return path")
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git master
testcase: will-it-scale
config: x86_64-rhel-9.4
compiler: gcc-12
test machine: 104 threads 2 sockets (Skylake) with 192G memory
parameters:
nr_task: 100%
mode: process
test: poll2
cpufreq_governor: performance
Details are as below:
-------------------------------------------------------------------------------------------------->
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20250612/202506121540.6eafcec4-lkp@xxxxxxxxx
=========================================================================================
compiler/cpufreq_governor/kconfig/mode/nr_task/rootfs/tbox_group/test/testcase:
gcc-12/performance/x86_64-rhel-9.4/process/100%/debian-12-x86_64-20240206.cgz/lkp-skl-fpga01/poll2/will-it-scale
commit:
f1745496d3 ("netfs: Update main API document")
5730609ffd ("select: do_pollfd: add unlikely branch hint return path")
f1745496d3fba34a 5730609ffd7e558e1e3305d0c68
---------------- ---------------------------
%stddev %change %stddev
\ | \
0.08 ± 31% -35.8% 0.05 ± 31% perf-sched.sch_delay.avg.ms.schedule_timeout.__wait_for_common.wait_for_completion_state.kernel_clone
2800 ± 91% +420.5% 14576 ±121% proc-vmstat.numa_hint_faults
1013 ± 37% +226.4% 3308 ±101% proc-vmstat.numa_hint_faults_local
25518332 +3.9% 26506874 will-it-scale.104.processes
245368 +3.9% 254873 will-it-scale.per_process_ops
25518332 +3.9% 26506874 will-it-scale.workload
4.802e+10 +3.8% 4.983e+10 perf-stat.i.branch-instructions
1.475e+08 +3.4% 1.525e+08 perf-stat.i.branch-misses
1.04 -3.6% 1.00 perf-stat.i.cpi
2.702e+11 +3.9% 2.808e+11 perf-stat.i.instructions
0.97 +3.7% 1.00 perf-stat.i.ipc
1.03 -3.6% 1.00 perf-stat.overall.cpi
0.97 +3.7% 1.00 perf-stat.overall.ipc
4.786e+10 +3.8% 4.966e+10 perf-stat.ps.branch-instructions
1.47e+08 +3.4% 1.52e+08 perf-stat.ps.branch-misses
2.693e+11 +3.9% 2.799e+11 perf-stat.ps.instructions
8.15e+13 +3.9% 8.468e+13 perf-stat.total.instructions
42.32 -4.1 38.22 perf-profile.calltrace.cycles-pp.fdget.do_poll.do_sys_poll.__x64_sys_poll.do_syscall_64
69.75 -2.1 67.63 perf-profile.calltrace.cycles-pp.do_sys_poll.__x64_sys_poll.do_syscall_64.entry_SYSCALL_64_after_hwframe.__poll
70.28 -2.1 68.17 perf-profile.calltrace.cycles-pp.__x64_sys_poll.do_syscall_64.entry_SYSCALL_64_after_hwframe.__poll
72.20 -2.0 70.23 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.__poll
76.46 -1.8 74.67 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.__poll
58.10 -1.5 56.63 perf-profile.calltrace.cycles-pp.do_poll.do_sys_poll.__x64_sys_poll.do_syscall_64.entry_SYSCALL_64_after_hwframe
94.49 -0.5 93.97 perf-profile.calltrace.cycles-pp.__poll
0.70 +0.0 0.72 perf-profile.calltrace.cycles-pp.__virt_addr_valid.check_heap_object.__check_object_size.do_sys_poll.__x64_sys_poll
0.92 +0.0 0.95 perf-profile.calltrace.cycles-pp.check_heap_object.__check_object_size.do_sys_poll.__x64_sys_poll.do_syscall_64
0.54 +0.0 0.58 ± 2% perf-profile.calltrace.cycles-pp.syscall_exit_to_user_mode.do_syscall_64.entry_SYSCALL_64_after_hwframe.__poll
1.94 +0.1 2.00 perf-profile.calltrace.cycles-pp.__check_object_size.do_sys_poll.__x64_sys_poll.do_syscall_64.entry_SYSCALL_64_after_hwframe
1.68 +0.1 1.74 perf-profile.calltrace.cycles-pp.rep_movs_alternative._copy_from_user.do_sys_poll.__x64_sys_poll.do_syscall_64
2.34 +0.1 2.40 perf-profile.calltrace.cycles-pp._copy_from_user.do_sys_poll.__x64_sys_poll.do_syscall_64.entry_SYSCALL_64_after_hwframe
0.89 +0.1 0.96 perf-profile.calltrace.cycles-pp.kfree.do_sys_poll.__x64_sys_poll.do_syscall_64.entry_SYSCALL_64_after_hwframe
2.85 +0.1 2.98 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64.__poll
5.86 +0.3 6.16 perf-profile.calltrace.cycles-pp.entry_SYSRETQ_unsafe_stack.__poll
7.26 +0.5 7.72 perf-profile.calltrace.cycles-pp.syscall_return_via_sysret.__poll
5.00 ± 2% +0.5 5.51 ± 2% perf-profile.calltrace.cycles-pp.testcase
2.27 ± 3% +0.6 2.84 ± 4% perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_safe_stack.__poll
42.25 -2.8 39.42 perf-profile.children.cycles-pp.fdget
58.21 -2.7 55.52 perf-profile.children.cycles-pp.do_poll
70.35 -2.1 68.24 perf-profile.children.cycles-pp.__x64_sys_poll
69.87 -2.1 67.77 perf-profile.children.cycles-pp.do_sys_poll
72.26 -2.0 70.29 perf-profile.children.cycles-pp.do_syscall_64
76.58 -1.8 74.80 perf-profile.children.cycles-pp.entry_SYSCALL_64_after_hwframe
95.09 -0.5 94.58 perf-profile.children.cycles-pp.__poll
0.71 +0.0 0.73 perf-profile.children.cycles-pp.__virt_addr_valid
0.99 +0.0 1.02 perf-profile.children.cycles-pp.check_heap_object
0.19 ± 2% +0.0 0.22 ± 3% perf-profile.children.cycles-pp.poll_freewait
0.54 +0.0 0.59 ± 2% perf-profile.children.cycles-pp.syscall_exit_to_user_mode
1.70 +0.1 1.76 perf-profile.children.cycles-pp.rep_movs_alternative
2.09 +0.1 2.14 perf-profile.children.cycles-pp.__check_object_size
0.89 +0.1 0.96 perf-profile.children.cycles-pp.kfree
2.56 +0.1 2.64 perf-profile.children.cycles-pp._copy_from_user
6.28 +0.3 6.60 perf-profile.children.cycles-pp.entry_SYSRETQ_unsafe_stack
1.30 ± 2% +0.3 1.61 ± 2% perf-profile.children.cycles-pp.entry_SYSCALL_64_safe_stack
3.84 +0.4 4.22 perf-profile.children.cycles-pp.entry_SYSCALL_64
7.33 +0.5 7.79 perf-profile.children.cycles-pp.syscall_return_via_sysret
5.01 ± 2% +0.5 5.52 ± 2% perf-profile.children.cycles-pp.testcase
40.83 -1.6 39.20 perf-profile.self.cycles-pp.fdget
17.11 -1.0 16.07 perf-profile.self.cycles-pp.do_poll
0.16 ± 3% +0.0 0.18 ± 3% perf-profile.self.cycles-pp.poll_freewait
0.65 +0.0 0.68 perf-profile.self.cycles-pp.__virt_addr_valid
0.98 +0.0 1.01 perf-profile.self.cycles-pp._copy_from_user
0.42 ± 2% +0.0 0.46 ± 2% perf-profile.self.cycles-pp.syscall_exit_to_user_mode
1.54 +0.0 1.58 perf-profile.self.cycles-pp.rep_movs_alternative
1.08 +0.1 1.14 ± 2% perf-profile.self.cycles-pp.__poll
0.88 +0.1 0.95 perf-profile.self.cycles-pp.kfree
4.39 ± 2% +0.2 4.59 ± 2% perf-profile.self.cycles-pp.entry_SYSCALL_64_after_hwframe
6.22 +0.3 6.53 perf-profile.self.cycles-pp.entry_SYSRETQ_unsafe_stack
3.41 +0.4 3.78 perf-profile.self.cycles-pp.entry_SYSCALL_64
7.32 +0.5 7.78 perf-profile.self.cycles-pp.syscall_return_via_sysret
4.82 ± 2% +0.5 5.32 ± 2% perf-profile.self.cycles-pp.testcase
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
