Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.kernel > #1727250

Re: Abysmal scheduler performance in Linus' tree?

From Andy Lutomirski <luto@amacapital.net>
Newsgroups linux.kernel
Subject Re: Abysmal scheduler performance in Linus' tree?
Date 2017-09-06 11:00 +0200
Message-ID <umElI-7Em-27@gated-at.bofh.it> (permalink)
References <umAUN-57m-1@gated-at.bofh.it> <umDSG-7ph-7@gated-at.bofh.it>
Organization linux.* mail to news gateway

Show all headers | View raw



> On Sep 6, 2017, at 1:25 AM, Peter Zijlstra <peterz@infradead.org> wrote:
> 
>> On Tue, Sep 05, 2017 at 10:13:39PM -0700, Andy Lutomirski wrote:
>> I'm running e7d0c41ecc2e372a81741a30894f556afec24315 from Linus' tree
>> today, and I'm seeing abysmal scheduler performance.  Running make -j4
>> ends up with all the tasks on CPU 3 most of the time (on my
>> 4-logical-thread laptop).  taskset -c 0 whatever puts whatever on CPU
>> 0, but plain while true; do true; done puts the infinite loop on CPU 3
>> right along with the make -j4 tasks.
>> 
>> This is on Fedora 26, and I don't think I'm doing anything weird.
>> systemd has enabled the cpu controller, but it doesn't seem to have
>> configured anything or created any non-root cgroups.
>> 
>> Just a heads up.  I haven't tried to diagnose it at all.
> 
> "make O=defconfig-build -j80" results in:
> 
> %Cpu0  : 90.7 us,  9.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu1  : 88.7 us, 11.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu2  : 93.5 us,  6.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu3  : 86.8 us, 13.2 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu4  : 89.7 us, 10.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu5  : 96.3 us,  3.7 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu6  : 95.3 us,  4.7 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu7  : 94.4 us,  5.6 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu8  : 91.7 us,  8.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu9  : 94.3 us,  5.7 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu10 : 90.7 us,  9.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu11 : 96.2 us,  3.8 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu12 : 91.5 us,  8.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu13 : 90.6 us,  9.4 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu14 : 97.2 us,  2.8 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu15 : 89.7 us, 10.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu16 : 90.6 us,  9.4 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu17 : 93.4 us,  6.6 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu18 : 90.6 us,  9.4 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu19 : 92.5 us,  7.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu20 : 94.4 us,  5.6 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu21 : 90.7 us,  9.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu22 : 92.5 us,  7.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu23 : 90.7 us,  9.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu24 : 91.6 us,  8.4 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu25 : 93.5 us,  6.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu26 : 93.4 us,  5.7 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.9 si,  0.0 st
> %Cpu27 : 92.5 us,  7.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu28 : 92.5 us,  7.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu29 : 88.8 us, 11.2 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu30 : 90.6 us,  9.4 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu31 : 93.5 us,  6.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu32 : 93.5 us,  6.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu33 : 93.4 us,  6.6 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu34 : 90.7 us,  9.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu35 : 93.5 us,  6.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu36 : 90.7 us,  9.3 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu37 : 97.2 us,  2.8 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu38 : 92.5 us,  7.5 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> %Cpu39 : 92.6 us,  7.4 sy,  0.0 ni,  0.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
> 
> Do you have a .config somewhere?

I'll attach tomorrow.  I'll also test in a VM.

> Are you running with the systemd? Is it
> creating cpu cgroups?

Yes systemd, no cgroups.

> 
> Any specifics on your setup?

On further fiddling, I only see this after a suspend and resume cycle.

Back to linux.kernel | Previous | NextPrevious in thread | Next in thread | Find similar | Unroll thread


Thread

Abysmal scheduler performance in Linus' tree? Andy Lutomirski <luto@kernel.org> - 2017-09-06 07:20 +0200
  Re: Abysmal scheduler performance in Linus' tree? Peter Zijlstra <peterz@infradead.org> - 2017-09-06 10:30 +0200
    Re: Abysmal scheduler performance in Linus' tree? Andy Lutomirski <luto@amacapital.net> - 2017-09-06 11:00 +0200
      Re: Abysmal scheduler performance in Linus' tree? Peter Zijlstra <peterz@infradead.org> - 2017-09-06 11:10 +0200
        Re: Abysmal scheduler performance in Linus' tree? Peter Zijlstra <peterz@infradead.org> - 2017-09-06 18:20 +0200
          Re: Abysmal scheduler performance in Linus' tree? Mike Galbraith <efault@gmx.de> - 2017-09-07 08:20 +0200
            Re: Abysmal scheduler performance in Linus' tree? Ingo Molnar <mingo@kernel.org> - 2017-09-07 09:40 +0200
            [PATCH] sched/cpuset/pm: Fix cpuset vs suspend-resume Peter Zijlstra <peterz@infradead.org> - 2017-09-07 11:20 +0200
              Re: [PATCH] sched/cpuset/pm: Fix cpuset vs suspend-resume Peter Zijlstra <peterz@infradead.org> - 2017-09-07 11:30 +0200
                Re: [PATCH] sched/cpuset/pm: Fix cpuset vs suspend-resume "Rafael J. Wysocki" <rjw@rjwysocki.net> - 2017-09-07 13:10 +0200
                Re: [PATCH] sched/cpuset/pm: Fix cpuset vs suspend-resume Tejun Heo <tj@kernel.org> - 2017-09-07 22:40 +0200
              [tip:sched/urgent] sched/cpuset/pm: Fix cpuset vs. suspend-resume  bugs tip-bot for Peter Zijlstra <tipbot@zytor.com> - 2017-09-07 12:40 +0200
  Re: Abysmal scheduler performance in Linus' tree? Chris Wilson <chris@chris-wilson.co.uk> - 2017-09-06 11:20 +0200
    Re: Abysmal scheduler performance in Linus' tree? Peter Zijlstra <peterz@infradead.org> - 2017-09-06 11:30 +0200
      Re: Abysmal scheduler performance in Linus' tree? Peter Zijlstra <peterz@infradead.org> - 2017-09-06 12:50 +0200
        Re: Abysmal scheduler performance in Linus' tree? Peter Zijlstra <peterz@infradead.org> - 2017-09-06 13:00 +0200
          [tip:sched/urgent] sched/fair: Fix wake_affine_llc() balancing  rules tip-bot for Peter Zijlstra <tipbot@zytor.com> - 2017-09-07 10:30 +0200

csiph-web