Re: [PATCH v1] sched: steer waking task to empty cfs_rq for better latencies

[Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>]

* Peter Zijlstra <peterz@infradead.org> [2012-04-24 19:09:14]:

> On Tue, 2012-04-24 at 18:58 +0200, Peter Zijlstra wrote:
> > On Tue, 2012-04-24 at 22:26 +0530, Srivatsa Vaddagiri wrote:
> > > Steer a waking task towards a cpu where its cgroup has zero tasks (in
> > > order to provide it better sleeper credits and hence reduce its wakeup
> > > latency). 
> > 
> > That's just vile.. pjt could you post your global vruntime stuff so
> > vatsa can have a go at that?
> 
> That is, you're playing a deficiency we should fix, not exploit.
> 
> Also, you do a second loop over all those cpus right after we've already
> iterated them..

The first loop doesn't necessarily iterate thr' all cpus (as its looking
for a core that is fully idle - and hence breaks once it finds a busy
sibling).

> furthermore, that 100%+ gain is still way insane, what else is broken?

?

I have tried most benchmarks that were recommended for this kind of
change. Let me know if you suggest any other benchmark ..

> Did you try those paravirt tlb-flush patches and other such weirdness?

I will try that next. But IMHO the benefit of reduced wakeup latencies
should be over and above any benefit we get from paravirtualization.

- vatsa