Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-15 Thread Peter Zijlstra
On Thu, 2013-02-14 at 13:42 +0530, Preeti U Murthy wrote: > Hi Peter,Alex, > If the eligible cpus happen to be all the cpus,then iterating over all > the > cpus for idlest would be much worse than iterating over sched domains > right? Depends, doing a domain walk generally gets you 2n cpus

Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-15 Thread Peter Zijlstra
On Thu, 2013-02-14 at 13:42 +0530, Preeti U Murthy wrote: Hi Peter,Alex, If the eligible cpus happen to be all the cpus,then iterating over all the cpus for idlest would be much worse than iterating over sched domains right? Depends, doing a domain walk generally gets you 2n cpus visited --

Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-14 Thread Alex Shi
On 02/14/2013 04:12 PM, Preeti U Murthy wrote: > **START PATCH* > > sched:Improve balancing for fork/exec/wake tasks Can you test the patch with hackbench and aim7 with 2000 threads? plus the patch in the tip:core/locking tree:

Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-14 Thread Preeti U Murthy
Hi everyone, On 02/12/2013 03:52 PM, Peter Zijlstra wrote: > On Thu, 2013-01-24 at 11:06 +0800, Alex Shi wrote: >> Guess the search cpu from bottom to up in domain tree come from >> commit 3dbd5342074a1e sched: multilevel sbe sbf, the purpose is >> balancing over tasks on all level domains. >> >>

Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-14 Thread Preeti U Murthy
Hi everyone, On 02/12/2013 03:52 PM, Peter Zijlstra wrote: On Thu, 2013-01-24 at 11:06 +0800, Alex Shi wrote: Guess the search cpu from bottom to up in domain tree come from commit 3dbd5342074a1e sched: multilevel sbe sbf, the purpose is balancing over tasks on all level domains. This

Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-14 Thread Alex Shi
On 02/14/2013 04:12 PM, Preeti U Murthy wrote: **START PATCH* sched:Improve balancing for fork/exec/wake tasks Can you test the patch with hackbench and aim7 with 2000 threads? plus the patch in the tip:core/locking tree:

Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-13 Thread Alex Shi
On 02/12/2013 06:22 PM, Peter Zijlstra wrote: > On Thu, 2013-01-24 at 11:06 +0800, Alex Shi wrote: >> Guess the search cpu from bottom to up in domain tree come from >> commit 3dbd5342074a1e sched: multilevel sbe sbf, the purpose is >> balancing over tasks on all level domains. >> >> This

Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-13 Thread Alex Shi
On 02/12/2013 06:22 PM, Peter Zijlstra wrote: On Thu, 2013-01-24 at 11:06 +0800, Alex Shi wrote: Guess the search cpu from bottom to up in domain tree come from commit 3dbd5342074a1e sched: multilevel sbe sbf, the purpose is balancing over tasks on all level domains. This balancing cost too

Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-12 Thread Peter Zijlstra
On Thu, 2013-01-24 at 11:06 +0800, Alex Shi wrote: > Guess the search cpu from bottom to up in domain tree come from > commit 3dbd5342074a1e sched: multilevel sbe sbf, the purpose is > balancing over tasks on all level domains. > > This balancing cost too much if there has many domain/groups in a

Re: [patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-02-12 Thread Peter Zijlstra
On Thu, 2013-01-24 at 11:06 +0800, Alex Shi wrote: Guess the search cpu from bottom to up in domain tree come from commit 3dbd5342074a1e sched: multilevel sbe sbf, the purpose is balancing over tasks on all level domains. This balancing cost too much if there has many domain/groups in a

[patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-01-23 Thread Alex Shi
Guess the search cpu from bottom to up in domain tree come from commit 3dbd5342074a1e sched: multilevel sbe sbf, the purpose is balancing over tasks on all level domains. This balancing cost too much if there has many domain/groups in a large system. If we remove this code, we will get quick

[patch v4 05/18] sched: quicker balancing on fork/exec/wake

2013-01-23 Thread Alex Shi
Guess the search cpu from bottom to up in domain tree come from commit 3dbd5342074a1e sched: multilevel sbe sbf, the purpose is balancing over tasks on all level domains. This balancing cost too much if there has many domain/groups in a large system. If we remove this code, we will get quick