Messages in this thread | | | Date | Fri, 16 Nov 2012 20:05:15 +0000 | From | Mel Gorman <> | Subject | Re: Benchmark results: "Enhanced NUMA scheduling with adaptive affinity" |
| |
On Fri, Nov 16, 2012 at 08:50:18PM +0100, Andrea Arcangeli wrote: > Hi, > > On Fri, Nov 16, 2012 at 02:14:28PM +0000, Mel Gorman wrote: > > With some shuffling the question on what to consider for merging > > becomes > > > > > > 1. TLB optimisation patches 1-3? Patches 1-3 > > I assume you mean simply reshuffling 33-35 as 1-3. >
Yes.
> > 2. Stats for migration? Patches 4-6 > > 3. Common NUMA infrastructure? Patches 7-21 > > 4. Basic fault-driven policy, stats, ratelimits Patches 22-35 > > > > Patches 36-43 are complete cabbage and should not be considered at this > > stage. It should be possible to build the placement policies and the > > scheduling decisions from schednuma, autonuma, some combination of the > > above or something completely different on top of patches 1-35. > > > > Peter, Ingo, Andrea? > > The patches 1-35 looks a great foundation so I think they'd be an > ideal candidate for a first upstream inclusion. >
Thanks.
-- Mel Gorman SUSE Labs
| |