On Fri, 2013-07-19 at 20:37 +0200, Peter Zijlstra wrote:
> On Thu, Jul 18, 2013 at 12:06:39PM -0700, Jason Low wrote:
>
> > N = 1
> > -
> > 19.21% reaim [k] __read_lock_failed
> > 14.79% reaim [k] mspin_lock
> > 12.19% reaim [k] __write_l
On Thu, Jul 18, 2013 at 12:06:39PM -0700, Jason Low wrote:
> N = 1
> -
> 19.21% reaim [k] __read_lock_failed
> 14.79% reaim [k] mspin_lock
> 12.19% reaim [k] __write_lock_failed
> 7.87% reaim [k] _raw_spin_lock
On Thu, 2013-07-18 at 07:59 -0400, Rik van Riel wrote:
> On 07/18/2013 05:32 AM, Peter Zijlstra wrote:
> > On Wed, Jul 17, 2013 at 09:02:24PM -0700, Jason Low wrote:
> >
> >> I ran a few AIM7 workloads for the 8 socket HT enabled case and I needed
> >> to set N to more than 20 in order to get the b
On Thu, 2013-07-18 at 17:42 +0530, Srikar Dronamraju wrote:
> > >
> > > idle_balance(u64 idle_duration)
> > > {
> > > u64 cost = 0;
> > >
> > > for_each_domain(sd) {
> > > if (cost + sd->cost > idle_duration/N)
> > > break;
> > >
> > > ...
> > >
> > > sd->cost = (sd->cost
* Peter Zijlstra [2013-07-18 14:35:31]:
> On Thu, Jul 18, 2013 at 05:45:46PM +0530, Srikar Dronamraju wrote:
> > We take locks if and only if we see imbalance and want to pull the
> > tasks.
> > However if the newly idle balance is not finding an imbalance then this
> > may not be an issue.
> >
On Thu, Jul 18, 2013 at 05:45:46PM +0530, Srikar Dronamraju wrote:
> We take locks if and only if we see imbalance and want to pull the
> tasks.
> However if the newly idle balance is not finding an imbalance then this
> may not be an issue.
>
> Probably /proc/schedstats will give a better picture
* Rik van Riel [2013-07-18 07:59:22]:
> On 07/18/2013 05:32 AM, Peter Zijlstra wrote:
> >On Wed, Jul 17, 2013 at 09:02:24PM -0700, Jason Low wrote:
> >
> >>I ran a few AIM7 workloads for the 8 socket HT enabled case and I needed
> >>to set N to more than 20 in order to get the big performance gai
> >
> > idle_balance(u64 idle_duration)
> > {
> > u64 cost = 0;
> >
> > for_each_domain(sd) {
> > if (cost + sd->cost > idle_duration/N)
> > break;
> >
> > ...
> >
> > sd->cost = (sd->cost + this_cost) / 2;
> > cost += this_cost;
> > }
> > }
> >
> > I would've initi
On 07/18/2013 05:32 AM, Peter Zijlstra wrote:
On Wed, Jul 17, 2013 at 09:02:24PM -0700, Jason Low wrote:
I ran a few AIM7 workloads for the 8 socket HT enabled case and I needed
to set N to more than 20 in order to get the big performance gains.
One thing that I thought of was to have N be bas
On Wed, Jul 17, 2013 at 09:02:24PM -0700, Jason Low wrote:
> I ran a few AIM7 workloads for the 8 socket HT enabled case and I needed
> to set N to more than 20 in order to get the big performance gains.
>
> One thing that I thought of was to have N be based on how often idle
> balance attempts d
On Wed, 2013-07-17 at 20:01 +0200, Peter Zijlstra wrote:
> On Wed, Jul 17, 2013 at 01:51:51PM -0400, Rik van Riel wrote:
> > On 07/17/2013 12:18 PM, Peter Zijlstra wrote:
>
> > >So the way I see things is that the only way newidle balance can slow down
> > >things is if it runs when we could have
On Wed, 2013-07-17 at 20:01 +0200, Peter Zijlstra wrote:
> On Wed, Jul 17, 2013 at 01:51:51PM -0400, Rik van Riel wrote:
> > On 07/17/2013 12:18 PM, Peter Zijlstra wrote:
>
> > >So the way I see things is that the only way newidle balance can slow down
> > >things is if it runs when we could have
On Wed, Jul 17, 2013 at 01:51:51PM -0400, Rik van Riel wrote:
> On 07/17/2013 12:18 PM, Peter Zijlstra wrote:
> >So the way I see things is that the only way newidle balance can slow down
> >things is if it runs when we could have ran something useful.
>
> Due to contention on the runqueue locks
On 07/17/2013 12:18 PM, Peter Zijlstra wrote:
On Wed, Jul 17, 2013 at 08:59:01AM -0700, Jason Low wrote:
Do you think its worth a try to consider each newidle balance attempt as
the total load_balance attempts until it is able to move a task, and
then skip balancing within the domain if a CP
On Wed, Jul 17, 2013 at 08:59:01AM -0700, Jason Low wrote:
>
> So if we have the following:
>
> for_each_domain(sd)
> before = sched_clock_cpu
> load_balance(sd)
> after = sched_clock_cpu
> idle_balance_completion_time = after - before
>
> At this point, the "idle_balanc
Hi Peter,
On Wed, 2013-07-17 at 11:39 +0200, Peter Zijlstra wrote:
> On Wed, Jul 17, 2013 at 01:11:41AM -0700, Jason Low wrote:
> > For the more complex model, are you suggesting that each completion time
> > is the time it takes to complete 1 iteration of the for_each_domain()
> > loop?
>
> Per
On Wed, Jul 17, 2013 at 01:11:41AM -0700, Jason Low wrote:
> For the more complex model, are you suggesting that each completion time
> is the time it takes to complete 1 iteration of the for_each_domain()
> loop?
Per sd, yes? So higher domains (or lower depending on how you model the thing
in yo
On Wed, 2013-07-17 at 09:25 +0200, Peter Zijlstra wrote:
> On Tue, Jul 16, 2013 at 03:48:01PM -0700, Jason Low wrote:
> > On Tue, 2013-07-16 at 22:20 +0200, Peter Zijlstra wrote:
> > > On Tue, Jul 16, 2013 at 12:21:03PM -0700, Jason Low wrote:
> > > > When running benchmarks on an 8 socket 80 core
On Wed, Jul 17, 2013 at 09:25:04AM +0200, Peter Zijlstra wrote:
> One thing that I thought of since is that we need to consider what
> happens for people with a low resolution sched_clock. IIRC there are
> still platforms that are jiffy based.
Ignore that, they're all UP.
--
To unsubscribe from th
On Tue, Jul 16, 2013 at 03:48:01PM -0700, Jason Low wrote:
> On Tue, 2013-07-16 at 22:20 +0200, Peter Zijlstra wrote:
> > On Tue, Jul 16, 2013 at 12:21:03PM -0700, Jason Low wrote:
> > > When running benchmarks on an 8 socket 80 core machine with a 3.10 kernel,
> > > there can be a lot of contentio
On Tue, 2013-07-16 at 22:20 +0200, Peter Zijlstra wrote:
> On Tue, Jul 16, 2013 at 12:21:03PM -0700, Jason Low wrote:
> > When running benchmarks on an 8 socket 80 core machine with a 3.10 kernel,
> > there can be a lot of contention in idle_balance() and related functions.
> > On many AIM7 workloa
On Tue, Jul 16, 2013 at 12:21:03PM -0700, Jason Low wrote:
> When running benchmarks on an 8 socket 80 core machine with a 3.10 kernel,
> there can be a lot of contention in idle_balance() and related functions.
> On many AIM7 workloads in which CPUs go idle very often and idle balance
> gets calle
On 07/16/2013 03:21 PM, Jason Low wrote:
When running benchmarks on an 8 socket 80 core machine with a 3.10 kernel,
there can be a lot of contention in idle_balance() and related functions.
On many AIM7 workloads in which CPUs go idle very often and idle balance
gets called a lot, it is actually
When running benchmarks on an 8 socket 80 core machine with a 3.10 kernel,
there can be a lot of contention in idle_balance() and related functions.
On many AIM7 workloads in which CPUs go idle very often and idle balance
gets called a lot, it is actually lowering performance.
Since idle balance o
24 matches
Mail list logo
