[<prev] [next>] [day] [month] [year] [list]
Message-ID: <202408042135.26c3555-oliver.sang@intel.com>
Date: Sun, 4 Aug 2024 21:45:30 +0800
From: kernel test robot <oliver.sang@...el.com>
To: Peter Zijlstra <peterz@...radead.org>
CC: <oe-lkp@...ts.linux.dev>, <lkp@...el.com>, Daniel Bristot de Oliveira
<bristot@...nel.org>, "Vineeth Pillai (Google)" <vineeth@...byteword.org>,
Juri Lelli <juri.lelli@...hat.com>, <linux-kernel@...r.kernel.org>,
<ying.huang@...el.com>, <feng.tang@...el.com>, <fengwei.yin@...el.com>,
<aubrey.li@...ux.intel.com>, <yu.c.chen@...el.com>, <oliver.sang@...el.com>
Subject: [peterz-queue:sched/prep] [sched/rt] 2cc53c6bfe:
rt-app.rt-app-thread2-1.wu_lat.average 25.9% regression
Hello,
kernel test robot noticed a 25.9% regression of rt-app.rt-app-thread2-1.wu_lat.average on:
commit: 2cc53c6bfe4bd6a39e9420e5b416e43344bb4482 ("sched/rt: Remove default bandwidth control")
https://git.kernel.org/cgit/linux/kernel/git/peterz/queue.git sched/prep
testcase: rt-app
test machine: 256 threads 2 sockets GENUINE INTEL(R) XEON(R) (Sierra Forest) with 128G memory
parameters:
workload: spreading-tasks
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@...el.com>
| Closes: https://lore.kernel.org/oe-lkp/202408042135.26c3555-oliver.sang@intel.com
Details are as below:
-------------------------------------------------------------------------------------------------->
The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20240804/202408042135.26c3555-oliver.sang@intel.com
=========================================================================================
compiler/cpufreq_governor/kconfig/rootfs/tbox_group/testcase/workload:
gcc-13/performance/x86_64-rhel-8.3/debian-12-x86_64-20240206.cgz/lkp-srf-2sp1/rt-app/spreading-tasks
commit:
add53a6969 ("sched/core: Fix picking of tasks for core scheduling with DL server")
2cc53c6bfe ("sched/rt: Remove default bandwidth control")
add53a696923bee0 2cc53c6bfe4bd6a39e9420e5b41
---------------- ---------------------------
%stddev %change %stddev
\ | \
0.03 ± 3% +0.0 0.04 ± 5% mpstat.cpu.all.irq%
0.01 ± 11% +0.0 0.01 ± 11% mpstat.cpu.all.soft%
950.00 -100.0% 0.00 sched_debug.rt_rq:.rt_runtime.avg
950.00 -100.0% 0.00 sched_debug.rt_rq:.rt_runtime.max
950.00 -100.0% 0.00 sched_debug.rt_rq:.rt_runtime.min
33.55 ± 16% +37.7% 46.21 ± 12% rt-app.rt-app-thread1-0.period.stddev
137.70 ± 9% +34.2% 184.83 ± 5% rt-app.rt-app-thread1-0.wu_lat.average
30.18 ± 5% +72.0% 51.89 ± 31% rt-app.rt-app-thread2-1.period.stddev
146.41 ± 2% +25.9% 184.29 ± 4% rt-app.rt-app-thread2-1.wu_lat.average
25.19 ± 9% +67.2% 42.11 ± 26% rt-app.rt-app-thread2-1.wu_lat.stddev
10482415 ± 3% +8.7% 11390605 perf-stat.i.cache-references
16774 ± 14% -19.1% 13565 ± 15% perf-stat.i.cycles-between-cache-misses
2.46 -4.5% 2.35 ± 2% perf-stat.i.ipc
0.12 ± 6% +12.5% 0.13 ± 9% perf-stat.overall.MPKI
2.30 -3.9% 2.21 ± 2% perf-stat.overall.ipc
10315907 ± 3% +8.6% 11200002 perf-stat.ps.cache-references
0.54 ± 45% +0.3 0.81 ± 18% perf-profile.calltrace.cycles-pp.sched_balance_find_dst_cpu.select_task_rq_fair.wake_up_new_task.kernel_clone.__do_sys_clone
0.54 ± 45% +0.3 0.81 ± 18% perf-profile.calltrace.cycles-pp.select_task_rq_fair.wake_up_new_task.kernel_clone.__do_sys_clone.do_syscall_64
0.45 ± 71% +0.3 0.79 ± 19% perf-profile.calltrace.cycles-pp.sched_balance_find_dst_group.sched_balance_find_dst_cpu.select_task_rq_fair.wake_up_new_task.kernel_clone
1.20 ± 28% +0.8 2.03 ± 19% perf-profile.calltrace.cycles-pp.update_process_times.tick_nohz_handler.__hrtimer_run_queues.hrtimer_interrupt.__sysvec_apic_timer_interrupt
1.31 ± 20% +1.2 2.49 ± 26% perf-profile.calltrace.cycles-pp.tick_nohz_handler.__hrtimer_run_queues.hrtimer_interrupt.__sysvec_apic_timer_interrupt.sysvec_apic_timer_interrupt
19.35 ± 23% -6.4 12.98 ± 22% perf-profile.children.cycles-pp.get_signal
19.40 ± 23% -6.3 13.12 ± 22% perf-profile.children.cycles-pp.arch_do_signal_or_restart
21.78 ± 18% -5.7 16.12 ± 14% perf-profile.children.cycles-pp.do_exit
21.78 ± 18% -5.7 16.12 ± 14% perf-profile.children.cycles-pp.do_group_exit
0.08 ±141% +0.2 0.24 ± 38% perf-profile.children.cycles-pp.dequeue_task_fair
0.22 ± 54% +0.3 0.50 ± 48% perf-profile.children.cycles-pp.__wp_page_copy_user
0.22 ± 54% +0.3 0.50 ± 48% perf-profile.children.cycles-pp.copy_mc_enhanced_fast_string
0.05 ±145% +0.5 0.60 ± 10% perf-profile.children.cycles-pp.update_rq_clock
0.22 ± 75% +0.6 0.84 ± 60% perf-profile.children.cycles-pp.flush_smp_call_function_queue
4.62 ± 17% +1.6 6.24 ± 10% perf-profile.children.cycles-pp.asm_sysvec_apic_timer_interrupt
0.22 ± 54% +0.3 0.50 ± 48% perf-profile.self.cycles-pp.copy_mc_enhanced_fast_string
0.03 ±223% +0.5 0.56 ± 15% perf-profile.self.cycles-pp.update_rq_clock
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
Powered by blists - more mailing lists