6:08 p.m.
Greeting,
FYI, we noticed a -27.9% regression of unixbench.score due to commit:
commit: faf71e54a3fdf0686194f5fcd772125deff9a397 ("vfs: Replace array of file
descriptors with an IDR")
git://git.infradead.org/users/willy/linux-dax.git xbitmap
in testcase: unixbench
on test machine: 24 threads Nehalem-EP with 24G memory
with following parameters:
runtime: 300s
nr_task: 1
test: syscall
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
In addition to that, the commit also has significant impact on the following tests:
+------------------+------------------------------------------------------------+
| testcase: change | unixbench: will-it-scale.per_process_ops -41.3% regression |
| test machine | qemu-system-x86_64 -enable-kvm -cpu host -smp 4 -m 5G |
| test parameters | cpufreq_governor=performance |
| | nr_task=100% |
| | runtime=300s |
| | test=syscall |
+------------------+------------------------------------------------------------+
| testcase: change | unixbench: unixbench.score 12.2% improvement |
| test machine | 8 threads Ivy Bridge with 16G memory |
| test parameters | cpufreq_governor=performance |
| | nr_task=100% |
| | runtime=300s |
| | test=syscall |
+------------------+------------------------------------------------------------+
Details are as below:
-------------------------------------------------------------------------------------------------->
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
testcase/path_params/tbox_group/run: unixbench/300s-1-syscall-performance/lkp-wsm-ep1
b11ed50f13fa0ef6 faf71e54a3fdf0686194f5fcd7
---------------- --------------------------
%stddev change %stddev
\ | \
2574 -28% 1857 unixbench.score
85.42 -25% 63.75 unixbench.time.user_time
214 10% 236 unixbench.time.system_time
0.06 -7% 0.06 ± 4% perf-stat.dTLB-load-miss-rate%
11164 ± 7% 25% 13920 ± 7% perf-stat.instructions-per-iTLB-miss
0.52 29% 0.67 ± 4% perf-stat.ipc
1.075e+12 ± 7% 25% 1.345e+12 ± 7% perf-stat.instructions
2.333e+11 23% 2.869e+11 perf-stat.branch-instructions
1.143e+12 33% 1.524e+12 perf-stat.iTLB-loads
2.053e+11 27% 2.611e+11 perf-stat.dTLB-stores
3.584e+11 11% 3.978e+11 perf-stat.dTLB-loads
22496651 -6% 21170258 ± 4% perf-stat.node-stores
0.01 -25% 0.01 perf-stat.iTLB-load-miss-rate%
0.34 -19% 0.28 perf-stat.branch-miss-rate%
0.08 -22% 0.06 perf-stat.dTLB-store-miss-rate%
perf-stat.dTLB-stores
3e+11 ++----------------------------------------------------------------+
| |
2.5e+11 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 O O
| |
*.*..*.*.*..*. .*.*.*..*.*.*..*.* *.*.*.. .*.*.. .*.*..*.*
2e+11 ++ *. : : *.*..* * |
| : : |
1.5e+11 ++ : : |
| : : |
1e+11 ++ : : |
| : : |
| : : |
5e+10 ++ :: |
| : |
0 ++---------------------------------*------------------------------+
perf-stat.iTLB-loads
1.8e+12 ++----------------------------------------------------------------+
| |
1.6e+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 O O
1.4e+12 ++ |
| .*. .*. |
1.2e+12 *+*..*.*.*..*.*..*.*.*..* *. * *.*.*..*.*..*.*.*..*.*.*..*.*
1e+12 ++ : : |
| : : |
8e+11 ++ : : |
6e+11 ++ : : |
| : : |
4e+11 ++ : : |
2e+11 ++ :: |
| : |
0 ++---------------------------------*------------------------------+
unixbench.score
3000 ++-------------------------------------------------------------------+
| |
2500 *+*..*.*..*.*..*.*..*.*..*.*..*.*..* *..*.*..*.*..*.*..*.*..*.*..*.*
| : : |
| : : |
2000 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 O O
| : : |
1500 ++ : : |
| : : |
1000 ++ : : |
| : : |
| :: |
500 ++ : |
| : |
0 ++-----------------------------------*-------------------------------+
unixbench.time.user_time
90 ++------------------*--*-----------------------------------------------+
*..*.*..*.*..*.*..* *..*.*..*..* *.*..*.*..*.*..*.*..*.*..*.*..*
80 ++ : : |
70 ++ : : |
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 O O
60 ++ : : |
50 ++ : : |
| : : |
40 ++ : : |
30 ++ : : |
| : : |
20 ++ :: |
10 ++ : |
| : |
0 ++------------------------------------*--------------------------------+
unixbench.time.system_time
250 ++--------------------------------------------------------------------+
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 O O
*..*.*..*.*..*.*..*.*..*.*..*.*..*.* *..*.*..*.*..*.*..*.*..*.*..*.*
200 ++ : : |
| : : |
| : : |
150 ++ : : |
| : : |
100 ++ : : |
| : : |
| : : |
50 ++ : : |
| : |
| : |
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