Greeting, FYI, we noticed a -7.4% regression of unixbench.score due to commit:

From kernel test robot <xiaolong.ye(a)intel.com&gt;:

commit: 625ed2bf049d5a352c1bcca962d6e133454eaaff ("sched/cfs: Make util/load_avg more stable") https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master in testcase: unixbench on test machine: 88 threads Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz with 64G memory with following parameters: runtime: 300s nr_task: 100% test: spawn cpufreq_governor: performance test-description: UnixBench is the original BYTE UNIX benchmark suite aims to test performance of Unix-like system. test-url: https://github.com/kdlucas/byte-unixbench Details are as below: --------------------------------------------------------------------------------------------------> To reproduce: git clone https://github.com/01org/lkp-tests.git cd lkp-tests bin/lkp install job.yaml # job file is attached in this email bin/lkp run job.yaml testcase/path_params/tbox_group/run: unixbench/300s-100%-spawn-performance/lkp-bdw-ep3b 8663effb24f94303 625ed2bf049d5a352c1bcca962 ---------------- -------------------------- %stddev change %stddev \ | \ 8888 -7% 8234 unixbench.score 11626 31% 15267 unixbench.time.system_time 5084 23% 6259 unixbench.time.percent_of_cpu_this_job_got 5203 5% 5455 unixbench.time.user_time 66039778 -7% 61588314 unixbench.time.voluntary_context_switches 7.932e+08 -7% 7.34e+08 unixbench.time.minor_page_faults 24502668 -52% 11794316 unixbench.time.involuntary_context_switches 628084 -17% 518637 interrupts.CAL:Function_call_interrupts 6000 ą 57% 1e+04 19033 ą 58% latency_stats.sum.call_rwsem_down_read_failed.__percpu_down_read.exit_signals.do_exit.do_group_exit.SyS_exit_group.entry_SYSCALL_64_fastpath 715117 ą 58% -4e+05 300172 ą 12% latency_stats.sum.io_schedule.__lock_page_or_retry.filemap_fault.__do_fault.__handle_mm_fault.handle_mm_fault.__do_page_fault.do_page_fault.page_fault 94622 96223 vmstat.system.in 500325 -16% 420024 vmstat.system.cs 1692 21% 2045 turbostat.Avg_MHz 60.71 21% 73.38 turbostat.%Busy 208 212 turbostat.PkgWatt 54.56 -8% 50.47 turbostat.RAMWatt 4.911e+13 21% 5.944e+13 perf-stat.cpu-cycles 6010 19% 7153 perf-stat.instructions-per-iTLB-miss 3.508e+12 14% 3.988e+12 perf-stat.branch-instructions 1.627e+13 10% 1.797e+13 perf-stat.instructions 4.504e+12 8% 4.886e+12 perf-stat.dTLB-loads 58.34 59.21 perf-stat.node-store-miss-rate% 42.85 42.00 perf-stat.iTLB-load-miss-rate% 3.609e+09 -4% 3.469e+09 perf-stat.iTLB-loads 2.125e+10 -5% 2.016e+10 perf-stat.branch-misses 2.707e+09 -7% 2.512e+09 perf-stat.iTLB-load-misses 7.939e+08 -7% 7.348e+08 perf-stat.page-faults 7.939e+08 -7% 7.348e+08 perf-stat.minor-faults 0.33 -9% 0.30 perf-stat.ipc 9.788e+09 -9% 8.927e+09 ą 3% perf-stat.dTLB-load-misses 14.74 -9% 13.43 perf-stat.cache-miss-rate% 3.426e+11 -9% 3.117e+11 perf-stat.cache-references 1.26e+09 -9% 1.141e+09 perf-stat.dTLB-store-misses 1.579e+12 -10% 1.421e+12 perf-stat.dTLB-stores 1.773e+10 -14% 1.523e+10 perf-stat.node-load-misses 5.685e+09 -15% 4.805e+09 perf-stat.node-store-misses 0.22 -16% 0.18 ą 3% perf-stat.dTLB-load-miss-rate% 1.666e+08 -16% 1.4e+08 perf-stat.context-switches 0.61 -17% 0.51 perf-stat.branch-miss-rate% 5.051e+10 -17% 4.187e+10 perf-stat.cache-misses 32471209 -18% 26608318 perf-stat.cpu-migrations 4.059e+09 -18% 3.311e+09 perf-stat.node-stores 8.13e+08 -24% 6.207e+08 perf-stat.node-loads unixbench.time.involuntary_context_switches 2.6e+07 ++----------------------------------------------------------------+ *.*.*.. .*.*.*. .*.*.*. .*..*.*.*. .*..*.*.*.* | 2.4e+07 ++ * *..*.*. .*.*. * * | 2.2e+07 ++ * | | | 2e+07 ++ | O O O O | 1.8e+07 ++O | | | 1.6e+07 ++ | 1.4e+07 ++ | | | 1.2e+07 ++ O O O O O O O O O O O O | | O O O O O O O O O O O O O O 1e+07 ++----------------------------------------------------------------+ perf-stat.cpu-cycles 6e+13 ++----------------------------------------------------------------+ | O O O O O O O O O O O O O O O O O O O O O O O O O O 5.8e+13 ++ | | | | | 5.6e+13 O+O O O | | O | 5.4e+13 ++ | | | 5.2e+13 ++ | | | | | 5e+13 ++ .*. .*.*.*..*.*.*.*. .*. .*. .*. | *.*.*. * *..*.*.* *..*.*.* *..*.* * | 4.8e+13 ++----------------------------------------------------------------+ perf-stat.node-load-misses 1.8e+10 ++---------------------------------------------------------------+ *.*.*..*.*.*.* .*.*..*.*.*.*.*.*..*.*.*.*.*.* | 1.75e+10 ++ : * | 1.7e+10 ++ : .*. + | | *.*. *.* | 1.65e+10 ++ | | | 1.6e+10 ++ | | | 1.55e+10 O+ O O O O O O O 1.5e+10 ++ O O O O O O O O | | | 1.45e+10 ++O O O O O O O O O O O O O | | O | 1.4e+10 ++---------------O-----------------------------------------------+ perf-stat.context-switches 1.7e+08 ++---------------------------------------------------------------+ *.*.*.. .*.*.* .*.*..*. .*.*.*.*.. .*.*.*.*.* | 1.65e+08 ++ * + * * * | 1.6e+08 ++ *. .*. + | | *. *.* | 1.55e+08 ++ | | | 1.5e+08 ++ | | | 1.45e+08 O+O O O O | 1.4e+08 ++ O O O O O O O O | | O O O O O O 1.35e+08 ++ O O O O O O O O O O | | O O | 1.3e+08 ++---------------------------------------------------------------+ perf-stat.cpu-migrations 3.3e+07 ++-----------------------------------------------------*-*--------+ | .*..*. .*. * .*.*. .*.*. + | 3.2e+07 *+* * *. + + .*..*.* *..*.* *..*.* | 3.1e+07 ++ *..*.* * | | | 3e+07 ++ | | | 2.9e+07 ++ | | | 2.8e+07 ++ | 2.7e+07 ++ | O O O O O O O O O O O O O 2.6e+07 ++ O O O O O O | | O O O O O O O O O O O O | 2.5e+07 ++----------------------------------------------------------------+ perf-stat.branch-miss-rate_ 0.62 ++-------------------------------------------------------------------+ *. .*. .*..*.*. *..*.*. .*. .*.*.. .*. .*.*.*.. .* | 0.6 ++*. * *.*.. + *. * * *. * | | *.* | 0.58 ++ | | | 0.56 ++ | | | 0.54 ++ | | | 0.52 ++ O O | O O O O O O | 0.5 ++ O O O O O O O O | O O O O O O O O O O | 0.48 ++O--O---O------O------O---------------------------------------------+ perf-stat.ipc 0.335 ++------------------------------------------------------------------+ *.*..*.*.*.*..* .*..*.*.*..*.*.*.*..*.*.*.*..*.* | 0.33 ++ : *.. * | 0.325 ++ : + + | | *.* *.* | 0.32 ++ | | | 0.315 ++ | | | 0.31 ++ | 0.305 ++ | O O O O O O O O O O O O O O 0.3 ++ O O O O O | | O O O O O O O O O O O O | 0.295 ++------------------------------------------------------------------+ perf-stat.instructions-per-iTLB-miss 7400 ++-------------------------------------------------------------------+ | O O O O O | 7200 ++O O O O O O O O O O O O O O O 7000 O+ O O O O O O O O O O | | | 6800 ++ | | | 6600 ++ | | | 6400 ++ | 6200 ++ *. .*. | *. + *..* *..*.*. .*.. | 6000 ++*..*. .*..*.* *..*. .* *.*. .*. .*..*.* | | * * *. * | 5800 ++-------------------------------------------------------------------+ [*] bisect-good sample [O] bisect-bad sample 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. Thanks, Xiaolong

kernel test robot <xiaolong.ye(a)intel.com&gt; writes: > Greeting, > > FYI, we noticed a -7.4% regression of unixbench.score due to commit: > > > commit: 625ed2bf049d5a352c1bcca962d6e133454eaaff ("sched/cfs: Make util/load_avg more stable") > https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master > > in testcase: unixbench > on test machine: 88 threads Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz with 64G memory > with following parameters: > > runtime: 300s > nr_task: 100% > test: spawn > cpufreq_governor: performance > > test-description: UnixBench is the original BYTE UNIX benchmark suite aims to test performance of Unix-like system. > This has been merged by v4.13-rc1, so we checked it again. If my understanding were correct, the patch changes the algorithm to calculate the load of CPU, so it influences the load balance behavior for this test case. 4.73 ± 8% -31.3% 3.25 ± 10% sched_debug.cpu.nr_running.max 0.95 ± 5% -29.0% 0.67 ± 4% sched_debug.cpu.nr_running.stddev As above, the effect is that the tasks are distributed into more CPUs, that is, system is more balanced. But this triggered more contention on tasklist_lock, so hurt the unixbench score, as below. 26.60 -10.6 16.05 perf-profile.calltrace.cycles-pp.intel_idle.cpuidle_enter_state.cpuidle_enter.call_cpuidle.do_idle 10.10 +2.4 12.53 perf-profile.calltrace.cycles-pp._raw_write_lock_irq.do_exit.do_group_exit.sys_exit_group.entry_SYSCALL_64_fastpath 8.03 +2.6 10.63 perf-profile.calltrace.cycles-pp._raw_write_lock_irq.release_task.wait_consider_task.do_wait.sys_wait4 17.98 +5.2 23.14 perf-profile.calltrace.cycles-pp._raw_read_lock.do_wait.sys_wait4.entry_SYSCALL_64_fastpath 7.47 +5.9 13.33 perf-profile.calltrace.cycles-pp._raw_write_lock_irq.copy_process._do_fork.sys_clone.do_syscall_64 The patch makes the tasks distributed more balanced, so I think scheduler do better job here. The problem is that the tasklist_lock isn't scalable. But considering this is only a micro-benchmark which specially exercises fork/exit/wait syscall, this may be not a big problem in reality. So, all in all, I think we can ignore this regression.