FYI, we noticed that vm-scalability.time.system_time -51.0% improvement on https://github.com/0day-ci/linux Kirill-A-Shutemov/thp-keep-huge-zero-page-pinned-until-tlb-flush/20160405-181706 commit 5f155ea5ab62b5321aa7315de62d21da59aa8c65 ("thp: keep huge zero page pinned until tlb flush") ========================================================================================= compiler/cpufreq_governor/kconfig/rootfs/runtime/tbox_group/test/testcase: gcc-4.9/performance/x86_64-rhel/debian-x86_64-2015-02-07.cgz/300s/lkp-hsx04/anon-r-seq/vm-scalability commit: v4.6-rc2 5f155ea5ab62b5321aa7315de62d21da59aa8c65 v4.6-rc2 5f155ea5ab62b5321aa7315de6 ---------------- -------------------------- %stddev %change %stddev \ | \ 517.69 ± 0% -51.0% 253.81 ± 0% vm-scalability.time.system_time 1694 ±173% +392.6% 8345 ±116% latency_stats.sum.load_module.SYSC_finit_module.SyS_finit_module.entry_SYSCALL_64_fastpath 517.69 ± 0% -51.0% 253.81 ± 0% time.system_time 0.46 ± 2% -16.5% 0.38 ± 2% turbostat.CPU%c6 1731 ± 14% +50.9% 2612 ± 9% cpuidle.C1-HSW.usage 15476193 ± 6% -32.9% 10389552 ± 6% cpuidle.C1E-HSW.time 3.55 ± 3% +26.0% 4.48 ± 1% perf-profile.cycles-pp.get_page_from_freelist.__alloc_pages_nodemask.alloc_pages_current.pte_alloc_one.do_huge_pmd_anonymous_page 2.45 ± 4% +9.9% 2.69 ± 5% perf-profile.cycles-pp.perf_event_task_tick.scheduler_tick.update_process_times.tick_sched_handle.tick_sched_timer 0.86 ± 5% +11.9% 0.96 ± 3% perf-profile.cycles-pp.update_curr.task_tick_fair.scheduler_tick.update_process_times.tick_sched_handle 28006 ±113% +143.3% 68134 ± 1% numa-vmstat.node0.nr_active_anon 1359 ± 44% +68.6% 2292 ± 28% numa-vmstat.node0.nr_mapped 82066 ± 31% -44.3% 45740 ± 60% numa-vmstat.node2.nr_file_pages 52443 ± 49% -69.2% 16160 ±171% numa-vmstat.node2.nr_shmem 336.60 ± 7% -20.5% 267.75 ± 8% slabinfo.kmem_cache.active_objs 336.60 ± 7% -20.5% 267.75 ± 8% slabinfo.kmem_cache.num_objs 694.40 ± 4% -12.4% 608.00 ± 4% slabinfo.kmem_cache_node.active_objs 742.40 ± 4% -11.6% 656.00 ± 4% slabinfo.kmem_cache_node.num_objs 136724 ± 93% +117.8% 297755 ± 1% numa-meminfo.node0.Active 111528 ±113% +144.5% 272658 ± 1% numa-meminfo.node0.Active(anon) 5429 ± 43% +70.6% 9262 ± 27% numa-meminfo.node0.Mapped 328531 ± 31% -44.3% 182976 ± 60% numa-meminfo.node2.FilePages 210042 ± 49% -69.2% 64658 ±171% numa-meminfo.node2.Shmem lkp-hsx04: Brickland Haswell-EX Memory: 512G vm-scalability.time.system_time 550 ++--------------------------------------------------------------------+ *..*..*..*.. .*..*..*..*.. .*.*..*..*.. | 500 ++ *..*..*.*..*..*..*..*..*. *. * | | | 450 ++ | | | 400 ++ | | | 350 ++ | | | 300 ++ | | O O | 250 O+ O O O O O O O O O O O O O O O O O O O O O O | | 200 ++--------------------------------------------------------------------+ time.system_time 550 ++--------------------------------------------------------------------+ *..*..*..*.. .*..*..*..*.. .*.*..*..*.. | 500 ++ *..*..*.*..*..*..*..*..*. *. * | | | 450 ++ | | | 400 ++ | | | 350 ++ | | | 300 ++ | | O O | 250 O+ O O O O O O O O O O O O O O O O O O O O O O | | 200 ++--------------------------------------------------------------------+ [*] bisect-good sample [O] bisect-bad sample To reproduce: git clone git://git.kernel.org/pub/scm/linux/kernel/git/wfg/lkp-tests.git cd lkp-tests bin/lkp install job.yaml # job file is attached in this email bin/lkp run job.yaml 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 Ye