Re: [lkp] [xfs] 68a9f5e700: aim7.jobs-per-min -13.6% regression

[Linus Torvalds]

On Tue, Aug 9, 2016 at 7:33 AM, kernel test robot <xiaolong.ye@xxxxxxxxx> wrote:
>
> FYI, we noticed a -13.6% regression of aim7.jobs-per-min due to commit:
> 68a9f5e7007c ("xfs: implement iomap based buffered write path")
>
> in testcase: aim7
> on test machine: 48 threads Ivytown Ivy Bridge-EP with 64G memory
> with following parameters:
>
>         disk: 1BRD_48G
>         fs: xfs
>         test: disk_wrt
>         load: 3000
>         cpufreq_governor: performance

Christop, Dave, was this expected?

>From looking at the numbers, it looks like much more IO going on (and
this less CPU load)..

>      37.23 Â  0%     +15.6%      43.04 Â  0%  aim7.time.elapsed_time
>      37.23 Â  0%     +15.6%      43.04 Â  0%  aim7.time.elapsed_time.max
>       6424 Â  1%     +31.3%       8432 Â  1%  aim7.time.involuntary_context_switches>       4003 Â  0%     +28.1%       5129 Â  1%  proc-vmstat.nr_active_file
>     979.25 Â  0%     +63.7%       1602 Â  1%  proc-vmstat.pgactivate
>       4699 Â  3%    +162.6%      12340 Â 73%  proc-vmstat.pgpgout
>      50.23 Â 19%     -27.3%      36.50 Â 17%  sched_debug.cpu.cpu_load[1].avg
>     466.50 Â 29%     -51.8%     225.00 Â 73%  sched_debug.cpu.cpu_load[1].max
>      77.78 Â 33%     -50.6%      38.40 Â 57%  sched_debug.cpu.cpu_load[1].stddev
>     300.50 Â 33%     -52.9%     141.50 Â 48%  sched_debug.cpu.cpu_load[2].max
>       1836 Â 10%     +65.5%       3039 Â  8%  slabinfo.scsi_data_buffer.active_objs
>       1836 Â 10%     +65.5%       3039 Â  8%  slabinfo.scsi_data_buffer.num_objs
>     431.75 Â 10%     +65.6%     715.00 Â  8%  slabinfo.xfs_efd_item.active_objs
>     431.75 Â 10%     +65.6%     715.00 Â  8%  slabinfo.xfs_efd_item.num_objs

but what do I know. Those profiles from the robot are pretty hard to
make sense of.

            Linus