On Tue, 2016-11-22 at 11:09 +0100, Peter Zijlstra wrote:
>
>
> Not a big fan of that part.. would not something like the below cure
> that?
>
> It would be slightly less optimal for Power7 but would actually be
> faster (on average) for the ITMT case, but most importantly, it does
> away with th
On Thu, Nov 10, 2016 at 02:36:33PM -0800, Tim Chen wrote:
> +/*
> + * Return the lowest numbered (or highest priority) cpu
> + * in the intersection of two cpu masks. If no cpu is
> + * is in both masks, nr_cpu_ids will be returned.
> + */
> +int __weak arch_asym_max_cpu_and(const struct cpumask *
We generalize the scheduler's asym packing to provide an ordering
of the cpu beyond just the cpu number. This allows the use of the
ASYM_PACKING scheduler machinery to move loads to preferred CPU in a
sched domain. The preference is defined with the cpu priority
given by arch_asym_cpu_priority(cpu
3 matches
Mail list logo
