Re: [PATCH v2] locking/osq_lock: Avoid false sharing in optimistic_spin_node

[kernel test robot]

Hello,

kernel test robot noticed a 15.3% improvement of fxmark.ssd_xfs_DWOM_72_bufferedio.works/sec on:


commit: 511ac0a137f4211f42aa2ba168e50550b703bb7c ("[PATCH v2] locking/osq_lock: Avoid false sharing in optimistic_spin_node")
url: https://github.com/intel-lab-lkp/linux/commits/Zeng-Heng/locking-osq_lock-Avoid-false-sharing-in-optimistic_spin_node/20231222-200921
base: https://git.kernel.org/cgit/linux/kernel/git/tip/tip.git a51749ab34d9e5dec548fe38ede7e01e8bb26454
patch link: https://lore.kernel.org/all/20231222121040.2635879-1-zengheng4@xxxxxxxxxx/
patch subject: [PATCH v2] locking/osq_lock: Avoid false sharing in optimistic_spin_node

testcase: fxmark
test machine: 128 threads 2 sockets Intel(R) Xeon(R) Platinum 8358 CPU @ 2.60GHz (Ice Lake) with 128G memory
parameters:

	disk: 1SSD
	media: ssd
	test: DWOM
	fstype: xfs
	directio: bufferedio
	cpufreq_governor: performance






Details are as below:
-------------------------------------------------------------------------------------------------->


The kernel config and materials to reproduce are available at:
https://download.01.org/0day-ci/archive/20240105/202401051804.77722270-oliver.sang@xxxxxxxxx

=========================================================================================
compiler/cpufreq_governor/directio/disk/fstype/kconfig/media/rootfs/tbox_group/test/testcase:
  gcc-12/performance/bufferedio/1SSD/xfs/x86_64-rhel-8.3/ssd/debian-11.1-x86_64-20220510.cgz/lkp-icl-2sp5/DWOM/fxmark

commit: 
  a51749ab34 ("locking/mutex: Document that mutex_unlock() is non-atomic")
  511ac0a137 ("locking/osq_lock: Avoid false sharing in optimistic_spin_node")

a51749ab34d9e5de 511ac0a137f4211f42aa2ba168e 
---------------- --------------------------- 
         %stddev     %change         %stddev
             \          |                \  
     19727 ±  6%     -13.5%      17073 ±  6%  vmstat.system.cs
     19254 ±  7%     -13.4%      16678 ±  7%  perf-stat.i.context-switches
     19499 ±  6%     -13.6%      16851 ±  6%  perf-stat.ps.context-switches
    105.12 ± 13%    +115.9%     227.00 ±  9%  perf-c2c.DRAM.local
      5028 ±  2%      -9.1%       4571 ±  2%  perf-c2c.DRAM.remote
      3526 ±  3%     -13.8%       3040 ±  3%  perf-c2c.HITM.remote
  77565522 ±  2%     +11.2%   86238276 ±  3%  fxmark.ssd_xfs_DWOM_18_bufferedio.works
   1551311 ±  2%     +11.2%    1724766 ±  3%  fxmark.ssd_xfs_DWOM_18_bufferedio.works/sec
  61955316           +12.3%   69564863        fxmark.ssd_xfs_DWOM_36_bufferedio.works
   1239107           +12.3%    1391298        fxmark.ssd_xfs_DWOM_36_bufferedio.works/sec
      2.21 ±  4%      +9.2%       2.41 ±  2%  fxmark.ssd_xfs_DWOM_4_bufferedio.user_sec
      1.10 ±  4%      +9.3%       1.20 ±  2%  fxmark.ssd_xfs_DWOM_4_bufferedio.user_util
  46595523           +16.2%   54146121 ±  2%  fxmark.ssd_xfs_DWOM_54_bufferedio.works
    931911           +16.2%    1082933 ±  2%  fxmark.ssd_xfs_DWOM_54_bufferedio.works/sec
  44518641           +15.3%   51332276        fxmark.ssd_xfs_DWOM_72_bufferedio.works
    890372           +15.3%    1026646        fxmark.ssd_xfs_DWOM_72_bufferedio.works/sec




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