Re: [PATCH v4 6/7] sched: add function nr_running_cpu to expose number of tasks running on cpu

[Mike Galbraith <umgwanakikbuti@gmail.com>]

On Tue, 2014-07-15 at 21:03 +0200, Peter Zijlstra wrote: 
> On Tue, Jul 15, 2014 at 08:06:55PM +0200, Mike Galbraith wrote:
> > On Tue, 2014-07-15 at 16:53 +0200, Peter Zijlstra wrote: 
> > > On Tue, Jul 15, 2014 at 04:45:25PM +0200, Mike Galbraith wrote:
> > > > 
> > > > 3.0.101-default        3.753363 usecs/loop -- avg 3.770737 530.4 KHz   1.000
> > > > 3.14.10-default        4.145348 usecs/loop -- avg 4.139987 483.1 KHz    .910     1.000
> > > > 3.15.4-default         4.355594 usecs/loop -- avg 4.351961 459.6 KHz    .866      .951    1.000
> > > > 3.16.0-default         4.537279 usecs/loop -- avg 4.543532 440.2 KHz    .829      .911     .957
> > > > 
> > > > 3.0.101-smp            3.692377 usecs/loop -- avg 3.690774 541.9 KHz   1.000
> > > > 3.14.10-smp            4.010009 usecs/loop -- avg 4.009019 498.9 KHz    .920
> > > > 3.15.4-smp             3.882398 usecs/loop -- avg 3.884095 514.9 KHz    .950
> > > > 3.16.0-master          4.061003 usecs/loop -- avg 4.058244 492.8 KHz    .909
> > > 
> > > Urgh,.. I need to go fix that :/
> > 
> > I'm poking about.  It's not just one thing 'course, just lots of change
> > adding up to less than wonderful.  Idle changes are costing some, for
> > obese config, avg goop.  The select_next_task() reorganization appears
> > to be costing, but eyeballing, I can see no excuse for that at all.
> 
> How is the idle stuff costing, cpu-affine pipe-test should pretty much
> peg a cpu at 100%, right? Or did I mis-understand and are you running a
> loose pipe-test?

Exactly, I'm measuring the cost of popping in and out of idle while
trying to reclaim overlap (which doesn't exist in pipe-test case).

-Mike