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

Groups > linux.kernel > #1440404 > unrolled thread

Re: [PATCH v2 04/13] sched: Introduce SD_ASYM_CPUCAPACITY sched_domain topology flag

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH v2 04/13] sched: Introduce SD_ASYM_CPUCAPACITY  sched_domain topology flag Peter Zijlstra <peterz@infradead.org> - 2016-07-11 12:00 +0200
    Re: [PATCH v2 04/13] sched: Introduce SD_ASYM_CPUCAPACITY  sched_domain topology flag Morten Rasmussen <morten.rasmussen@arm.com> - 2016-07-11 12:50 +0200

#1440404 — Re: [PATCH v2 04/13] sched: Introduce SD_ASYM_CPUCAPACITY sched_domain topology flag

On Wed, Jun 22, 2016 at 06:03:15PM +0100, Morten Rasmussen wrote:
> Add a topology flag to the sched_domain hierarchy indicating
> sched_groups at this sched_domain level having different per cpu
> capacity (e.g. big.LITTLE big-only and little-only groups) or groups in
> sibling domains with different capacity.

>                                          IOW, domains not spanning cpus
> of all available capacities up until and including the first level
> spanning cpus of all capacities available system wide.

This ^, which would be SD1 below, why? I would expect only SD2 to be
tagged, since that does indeed have asymmetric capacities.

> 
> SD 2      [  0      1      2      3]  SD_ASYM_CPUCAPACITY
> 
> SD 1      [  0      1] [   2      3]  SD_ASYM_CPUCAPACITY
>
> cpu:         0      1      2      3
> capacity:  756    756   1024   1024
> 

[toc] | [next] | [standalone]

#1440464

On Mon, Jul 11, 2016 at 11:55:23AM +0200, Peter Zijlstra wrote:
> On Wed, Jun 22, 2016 at 06:03:15PM +0100, Morten Rasmussen wrote:
> > Add a topology flag to the sched_domain hierarchy indicating
> > sched_groups at this sched_domain level having different per cpu
> > capacity (e.g. big.LITTLE big-only and little-only groups) or groups in
> > sibling domains with different capacity.
> 
> >                                          IOW, domains not spanning cpus
> > of all available capacities up until and including the first level
> > spanning cpus of all capacities available system wide.
> 
> This ^, which would be SD1 below, why? I would expect only SD2 to be
> tagged, since that does indeed have asymmetric capacities.

That would make most sense, but it complicates how we can use the flag.
See reply to your related question for patch #5.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web