[norov:kerne-unload] [tracing] a28f9a04b2: will-it-scale.per_process_ops 16.3% improvement
kernel test robot <[email protected]>
| Newsgroups | dev.linux.lists.oe-lkp |
|---|---|
| Message-ID | <[email protected]> |
Hello,
kernel test robot noticed a 16.3% improvement of will-it-scale.per_process_ops on:
commit: a28f9a04b26374e653258db4f852469961dfdb68 ("tracing: Remove size parameter in __trace_puts()")
https://github.com/norov/linux kerne-unload
testcase: will-it-scale
config: x86_64-rhel-9.4
compiler: gcc-14
test machine: 224 threads 4 sockets Intel(R) Xeon(R) Platinum 8380H CPU @ 2.90GHz (Cooper Lake) with 192G memory
parameters:
nr_task: 100%
mode: process
test: lseek1
cpufreq_governor: performance
Details are as below:
-------------------------------------------------------------------------------------------------->
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20251229/[email protected]
=========================================================================================
compiler/cpufreq_governor/kconfig/mode/nr_task/rootfs/tbox_group/test/testcase:
gcc-14/performance/x86_64-rhel-9.4/process/100%/debian-13-x86_64-20250902.cgz/lkp-cpl-4sp2/lseek1/will-it-scale
commit:
12c3bbf898 ("kernel.h: include linux/instruction_pointer.h explicitly")
a28f9a04b2 ("tracing: Remove size parameter in __trace_puts()")
12c3bbf89800653f a28f9a04b26374e653258db4f85
---------------- ---------------------------
%stddev %change %stddev
\ | \
24.86 +3.0 27.84 mpstat.cpu.all.usr%
0.14 ± 6% -13.0% 0.12 ± 9% sched_debug.cfs_rq:/.h_nr_runnable.stddev
0.50 +16.9% 0.59 turbostat.IPC
36479 ± 75% +143.7% 88884 ± 21% numa-vmstat.node1.nr_active_anon
36479 ± 75% +143.7% 88884 ± 21% numa-vmstat.node1.nr_zone_active_anon
21982 ± 32% -47.0% 11653 ± 12% numa-vmstat.node3.nr_mapped
800.17 ± 6% -22.6% 619.67 ± 14% perf-c2c.DRAM.remote
1142 ± 8% -47.5% 600.33 ± 14% perf-c2c.HITM.local
710.00 ± 8% -21.9% 554.50 ± 15% perf-c2c.HITM.remote
1.332e+09 +16.3% 1.549e+09 will-it-scale.224.processes
5947747 +16.3% 6914856 will-it-scale.per_process_ops
1.332e+09 +16.3% 1.549e+09 will-it-scale.workload
146159 ± 75% +143.3% 355653 ± 21% numa-meminfo.node1.Active
145938 ± 75% +143.5% 355428 ± 22% numa-meminfo.node1.Active(anon)
761026 ± 26% +221.4% 2445772 ± 60% numa-meminfo.node1.MemUsed
88069 ± 32% -47.8% 45967 ± 13% numa-meminfo.node3.Mapped
0.01 ± 2% -11.5% 0.01 ± 3% perf-stat.i.MPKI
1.915e+11 +16.2% 2.226e+11 perf-stat.i.branch-instructions
3.414e+08 +14.7% 3.917e+08 perf-stat.i.branch-misses
10.14 ± 6% +3.0 13.12 ± 3% perf-stat.i.cache-miss-rate%
1730502 ± 4% -11.3% 1535699 ± 3% perf-stat.i.cache-misses
17825894 ± 6% -26.7% 13057672 ± 2% perf-stat.i.cache-references
1.32 -14.2% 1.13 perf-stat.i.cpi
654783 ± 5% +24.7% 816265 perf-stat.i.cycles-between-cache-misses
5.859e+11 +16.2% 6.808e+11 perf-stat.i.instructions
0.76 +16.6% 0.89 perf-stat.i.ipc
0.35 ± 4% -15.5% 0.29 ± 9% perf-stat.i.major-faults
0.00 ± 4% -23.6% 0.00 ± 3% perf-stat.overall.MPKI
9.72 ± 4% +2.0 11.76 ± 3% perf-stat.overall.cache-miss-rate%
1.32 -14.3% 1.13 perf-stat.overall.cpi
446928 ± 4% +12.1% 500993 ± 3% perf-stat.overall.cycles-between-cache-misses
0.76 +16.7% 0.89 perf-stat.overall.ipc
1.909e+11 +16.2% 2.218e+11 perf-stat.ps.branch-instructions
3.403e+08 +14.7% 3.904e+08 perf-stat.ps.branch-misses
1724803 ± 4% -11.2% 1531111 ± 3% perf-stat.ps.cache-misses
17787512 ± 6% -26.8% 13028723 ± 2% perf-stat.ps.cache-references
5.84e+11 +16.2% 6.785e+11 perf-stat.ps.instructions
0.34 ± 4% -15.4% 0.29 ± 9% perf-stat.ps.major-faults
1.765e+14 +16.1% 2.05e+14 perf-stat.total.instructions
26.52 -6.5 20.02 perf-profile.calltrace.cycles-pp.clear_bhb_loop.llseek
2.43 -0.4 2.04 perf-profile.calltrace.cycles-pp.x64_sys_call.do_syscall_64.entry_SYSCALL_64_after_hwframe.llseek
3.67 -0.3 3.42 perf-profile.calltrace.cycles-pp.syscall_return_via_sysret.llseek
1.63 -0.1 1.57 ± 2% perf-profile.calltrace.cycles-pp.generic_file_llseek_size.__x64_sys_lseek.do_syscall_64.entry_SYSCALL_64_after_hwframe.llseek
0.73 ± 3% +0.1 0.81 ± 2% perf-profile.calltrace.cycles-pp.testcase
98.60 +0.1 98.68 perf-profile.calltrace.cycles-pp.llseek
2.51 +0.1 2.64 perf-profile.calltrace.cycles-pp.shmem_file_llseek.__x64_sys_lseek.do_syscall_64.entry_SYSCALL_64_after_hwframe.llseek
0.82 +0.5 1.30 ± 18% perf-profile.calltrace.cycles-pp.entry_SYSRETQ_unsafe_stack.llseek
31.77 +0.7 32.48 perf-profile.calltrace.cycles-pp.do_syscall_64.entry_SYSCALL_64_after_hwframe.llseek
6.18 +1.0 7.20 perf-profile.calltrace.cycles-pp.arch_exit_to_user_mode_prepare.do_syscall_64.entry_SYSCALL_64_after_hwframe.llseek
6.70 +1.0 7.74 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_safe_stack.llseek
7.50 +1.1 8.60 perf-profile.calltrace.cycles-pp.fdget_pos.__x64_sys_lseek.do_syscall_64.entry_SYSCALL_64_after_hwframe.llseek
35.81 +1.2 37.04 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64_after_hwframe.llseek
24.16 +3.8 28.00 perf-profile.calltrace.cycles-pp.entry_SYSCALL_64.llseek
26.74 -6.4 20.31 perf-profile.children.cycles-pp.clear_bhb_loop
2.79 -0.4 2.36 perf-profile.children.cycles-pp.x64_sys_call
4.11 -0.2 3.92 perf-profile.children.cycles-pp.syscall_return_via_sysret
0.90 -0.1 0.82 perf-profile.children.cycles-pp.its_return_thunk
1.64 -0.1 1.58 ± 2% perf-profile.children.cycles-pp.generic_file_llseek_size
0.61 ± 2% -0.1 0.56 perf-profile.children.cycles-pp.lseek@plt
2.69 +0.2 2.85 perf-profile.children.cycles-pp.shmem_file_llseek
32.18 +0.7 32.92 perf-profile.children.cycles-pp.do_syscall_64
4.56 +0.7 5.30 perf-profile.children.cycles-pp.entry_SYSCALL_64_safe_stack
6.23 +1.0 7.28 perf-profile.children.cycles-pp.arch_exit_to_user_mode_prepare
7.50 +1.1 8.61 perf-profile.children.cycles-pp.fdget_pos
36.37 +1.4 37.72 perf-profile.children.cycles-pp.entry_SYSCALL_64_after_hwframe
9.20 +1.7 10.90 perf-profile.children.cycles-pp.entry_SYSRETQ_unsafe_stack
17.66 +2.7 20.40 perf-profile.children.cycles-pp.entry_SYSCALL_64
26.55 -6.5 20.08 perf-profile.self.cycles-pp.clear_bhb_loop
3.81 -1.2 2.61 ± 2% perf-profile.self.cycles-pp.__x64_sys_lseek
2.62 -0.4 2.20 perf-profile.self.cycles-pp.x64_sys_call
4.11 -0.2 3.91 perf-profile.self.cycles-pp.syscall_return_via_sysret
0.82 -0.1 0.71 perf-profile.self.cycles-pp.its_return_thunk
1.46 -0.1 1.36 ± 2% perf-profile.self.cycles-pp.generic_file_llseek_size
7.15 +0.1 7.25 perf-profile.self.cycles-pp.do_syscall_64
2.68 +0.2 2.84 perf-profile.self.cycles-pp.shmem_file_llseek
2.37 +0.4 2.79 perf-profile.self.cycles-pp.entry_SYSCALL_64_safe_stack
3.92 +0.7 4.61 perf-profile.self.cycles-pp.entry_SYSCALL_64_after_hwframe
6.04 +1.0 7.05 perf-profile.self.cycles-pp.arch_exit_to_user_mode_prepare
7.28 +1.2 8.43 perf-profile.self.cycles-pp.fdget_pos
8.97 +1.3 10.28 perf-profile.self.cycles-pp.entry_SYSCALL_64
8.76 +1.6 10.40 perf-profile.self.cycles-pp.entry_SYSRETQ_unsafe_stack
12.00 +2.0 14.02 perf-profile.self.cycles-pp.llseek
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