Greeting, FYI, we noticed a -16.2% regression of netperf.Throughput_tps due to commit: commit: 90001d67be2fa2acbe3510d1f64fa6533efa30ef ("sched/fair: Fix wake_affine() for !NUMA_BALANCING") https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master in testcase: netperf on test machine: 8 threads Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz with 8G memory with following parameters: ip: ipv4 runtime: 300s nr_threads: 200% cluster: cs-localhost test: TCP_RR cpufreq_governor: performance test-description: Netperf is a benchmark that can be use to measure various aspect of networking performance. test-url: http://www.netperf.org/netperf/ 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: netperf/ipv4-300s-200%-cs-localhost-TCP_RR-performance/lkp-hsw-d01 20435d84e5f2041c 90001d67be2fa2acbe3510d1f6 ---------------- -------------------------- 40233 -16% 33701 netperf.Throughput_tps 1.91e+08 -16% 1.609e+08 netperf.time.voluntary_context_switches 1096 1075 netperf.time.system_time 395 -4% 381 netperf.time.percent_of_cpu_this_job_got 95.58 -23% 73.89 netperf.time.user_time 22125 21748 interrupts.CAL:Function_call_interrupts 21198 20% 25506 vmstat.system.in 1295567 -8% 1192531 vmstat.system.cs 2026 -10% 1830 perf-stat.instructions-per-iTLB-miss 1.546e+10 -6% 1.459e+10 perf-stat.branch-misses 5.359e+10 114% 1.145e+11 perf-stat.cache-references 2.411e+10 -10% 2.16e+10 perf-stat.iTLB-loads 1.106e+12 -14% 9.496e+11 perf-stat.dTLB-stores 5842994 ± 9% 30% 7571350 ± 3% perf-stat.cache-misses 1.55 14% 1.78 perf-stat.cpi 724073 ± 3% 34% 968098 ± 3% perf-stat.node-stores 1.46 9% 1.59 perf-stat.branch-miss-rate% 0.64 -13% 0.56 perf-stat.ipc 0.01 ± 10% -39% 0.01 ± 4% perf-stat.cache-miss-rate% 1.06e+12 -14% 9.171e+11 perf-stat.branch-instructions 8.826e+12 8.715e+12 perf-stat.cpu-cycles 1.699e+12 -14% 1.456e+12 perf-stat.dTLB-loads 5.686e+12 -14% 4.905e+12 perf-stat.instructions 1.417e+09 -13% 1.228e+09 perf-stat.dTLB-store-misses 0.19 29% 0.25 ± 4% perf-stat.dTLB-load-miss-rate% 3.296e+09 10% 3.632e+09 ± 4% perf-stat.dTLB-load-misses 10.43 6% 11.04 perf-stat.iTLB-load-miss-rate% 40164 ± 23% 66526% 26759987 perf-stat.cpu-migrations 3731528 ± 4% 32% 4923150 ± 3% perf-stat.node-loads 2.807e+09 -5% 2.68e+09 perf-stat.iTLB-load-misses 3.935e+08 -8% 3.613e+08 perf-stat.context-switches netperf.Throughput_tps 43000 ++------------------------------------------------------------------+ 42000 ++ .*..*.*..*.* | | .*.*. + | 41000 ++ *. + .*..*.*.. | 40000 ++ + *..* *..*.*..*..* | | + | 39000 *+.*.*..*.* | 38000 ++ | 37000 ++ | | | 36000 ++ | 35000 ++ | | O O O O O O O O O | 34000 O+ O O O O O O O O O O O O O O O 33000 ++----------------O--O----------------------------------------------+ perf-stat.cpu-cycles 8.84e+12 ++---------------------------------------------------------------+ *..*.*..*.*..*.*..*.*..*.*..*.*..*.*..*.*..*. .*.*..*.* | 8.82e+12 ++ *. | | | 8.8e+12 ++ | | | 8.78e+12 ++ | | | 8.76e+12 ++ | | | 8.74e+12 ++ | | | 8.72e+12 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 O 8.7e+12 ++---------------------------------------------------------------+ perf-stat.cache-references 1.3e+11 ++----------------------------------------------------------------+ | O O O O O O O O O O | 1.2e+11 O+ O O O O O O O O O 1.1e+11 ++ O O O O O O O | | | 1e+11 ++ | 9e+10 ++ | | | 8e+10 ++ | 7e+10 ++ | *..*. .* | 6e+10 ++ *..* + | 5e+10 ++ + .*..*.*.. .*..*.*..*.*..*.*..*.*..* | | * *.*..*.*. | 4e+10 ++----------------------------------------------------------------+ perf-stat.cpu-migrations 3e+07 ++----------------------------------------------------------------+ | O O O O O O O O O O | 2.5e+07 O+ O O O O O O O O O O O O O O O O | | | | 2e+07 ++ | | | 1.5e+07 ++ | | | 1e+07 ++ | | | | | 5e+06 ++ | | | 0 *+-*-*--*-*--*-*--*-*--*-*--*-*--*--*-*--*-*--*-*--*-*--*---------+ [*] 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