Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Peter Zijlstra
On Tue, Feb 13, 2018 at 03:00:20PM +, Mel Gorman wrote: > On Tue, Feb 13, 2018 at 03:43:26PM +0100, Peter Zijlstra wrote: > > > > > > Well, it was deliberate. While it's possible to be on the same memory > > > node and not sharing cache, the scheduler typically is more concerned with > > >

Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Peter Zijlstra
On Tue, Feb 13, 2018 at 03:00:20PM +, Mel Gorman wrote: > On Tue, Feb 13, 2018 at 03:43:26PM +0100, Peter Zijlstra wrote: > > > > > > Well, it was deliberate. While it's possible to be on the same memory > > > node and not sharing cache, the scheduler typically is more concerned with > > >

Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Mel Gorman
On Tue, Feb 13, 2018 at 03:43:26PM +0100, Peter Zijlstra wrote: > > > > Well, it was deliberate. While it's possible to be on the same memory > > node and not sharing cache, the scheduler typically is more concerned with > > the LLC than NUMA per-se. If they share LLC, then I also assume that

Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Mel Gorman
On Tue, Feb 13, 2018 at 03:43:26PM +0100, Peter Zijlstra wrote: > > > > Well, it was deliberate. While it's possible to be on the same memory > > node and not sharing cache, the scheduler typically is more concerned with > > the LLC than NUMA per-se. If they share LLC, then I also assume that

Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Peter Zijlstra
On Tue, Feb 13, 2018 at 02:18:12PM +, Mel Gorman wrote: > On Tue, Feb 13, 2018 at 03:01:37PM +0100, Peter Zijlstra wrote: > > On Tue, Feb 13, 2018 at 01:37:30PM +, Mel Gorman wrote: > > > +static void > > > +update_wa_numa_placement(struct task_struct *p, int prev_cpu, int target) > > > +{

Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Peter Zijlstra
On Tue, Feb 13, 2018 at 02:18:12PM +, Mel Gorman wrote: > On Tue, Feb 13, 2018 at 03:01:37PM +0100, Peter Zijlstra wrote: > > On Tue, Feb 13, 2018 at 01:37:30PM +, Mel Gorman wrote: > > > +static void > > > +update_wa_numa_placement(struct task_struct *p, int prev_cpu, int target) > > > +{

Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Mel Gorman
On Tue, Feb 13, 2018 at 03:01:37PM +0100, Peter Zijlstra wrote: > On Tue, Feb 13, 2018 at 01:37:30PM +, Mel Gorman wrote: > > +static void > > +update_wa_numa_placement(struct task_struct *p, int prev_cpu, int target) > > +{ > > + unsigned long interval; > > + > > + if

Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Mel Gorman
On Tue, Feb 13, 2018 at 03:01:37PM +0100, Peter Zijlstra wrote: > On Tue, Feb 13, 2018 at 01:37:30PM +, Mel Gorman wrote: > > +static void > > +update_wa_numa_placement(struct task_struct *p, int prev_cpu, int target) > > +{ > > + unsigned long interval; > > + > > + if

Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Peter Zijlstra
On Tue, Feb 13, 2018 at 01:37:30PM +, Mel Gorman wrote: > +static void > +update_wa_numa_placement(struct task_struct *p, int prev_cpu, int target) > +{ > + unsigned long interval; > + > + if (!static_branch_likely(_numa_balancing)) > + return; > + > + /* If balancing

Re: [PATCH 6/6] sched/numa: Delay retrying placement for automatic NUMA balance after wake_affine

2018-02-13 Thread Peter Zijlstra
On Tue, Feb 13, 2018 at 01:37:30PM +, Mel Gorman wrote: > +static void > +update_wa_numa_placement(struct task_struct *p, int prev_cpu, int target) > +{ > + unsigned long interval; > + > + if (!static_branch_likely(_numa_balancing)) > + return; > + > + /* If balancing