On Mon, 1 Mar 2010, Mel Gorman wrote:
> Christoph, how feasible would it be to allow parallel reclaimers in
> __zone_reclaim() that back off at a rate depending on the number of
> reclaimers?
Not too hard. Zone locking is there but there may be a lot of bouncing
cachelines if you run it concurren
On Tue, Feb 23, 2010 at 12:55:51PM +1100, Anton Blanchard wrote:
>
> Hi Mel,
>
I'm back but a bit vague. Am on painkillers for the bashing I gave
myself down the hills.
> > You're pretty much on the button here. Only one thread at a time enters
> > zone_reclaim. The others back off and try the
On Tue, 23 Feb 2010, Anton Blanchard wrote:
> zone_reclaim_mode.txt
> Now we set zone_reclaim_mode = 1. On each iteration we continue to improve,
> but even after 10 runs of stream we have > 10% remote node memory usage.
The intend of zone reclaim was never to allocate all memory from on node.
Yo
On Tue, Feb 23, 2010 at 12:55:51PM +1100, Anton Blanchard wrote:
>
> Hi Mel,
>
I'm afraid I'm on vacation at the moment. This mail is costing me shots with
penaltys every minute it's open. It'll be early next week before I can look
at this closely.
Sorry.
> > You're pretty much on the button
Hi Mel,
> You're pretty much on the button here. Only one thread at a time enters
> zone_reclaim. The others back off and try the next zone in the zonelist
> instead. I'm not sure what the original intention was but most likely it
> was to prevent too many parallel reclaimers in the same zone po
Hi Balbir,
> A reclaim distance of 10 implies a ratio of 1, that means we'll always
> do zone_reclaim() to free page cache and slab cache before moving on
> to another node?
I want to make an effort to reclaim local pagecache before ever going
off node. As an example, a completely off node stre
* Christoph Lameter [2010-02-19 09:51:12]:
> On Fri, 19 Feb 2010, Balbir Singh wrote:
>
> > >> zone_reclaim. The others back off and try the next zone in the zonelist
> > >> instead. I'm not sure what the original intention was but most likely it
> > >> was to prevent too many parallel reclaimer
On Fri, 19 Feb 2010, Balbir Singh wrote:
> >> zone_reclaim. The others back off and try the next zone in the zonelist
> >> instead. I'm not sure what the original intention was but most likely it
> >> was to prevent too many parallel reclaimers in the same zone potentially
> >> dumping out way mor
On Fri, Feb 19, 2010 at 3:59 AM, Anton Blanchard wrote:
>
> I noticed /proc/sys/vm/zone_reclaim_mode was 0 on a ppc64 NUMA box. It gets
> enabled via this:
>
> /*
> * If another node is sufficiently far away then it is better
> * to reclaim pages in a zone before going off n
On Fri, Feb 19, 2010 at 8:42 PM, Christoph Lameter
wrote:
> On Fri, 19 Feb 2010, Mel Gorman wrote:
>
>> > > The patch below sets a smaller value for RECLAIM_DISTANCE and thus
>> > > enables
>> > > zone reclaim.
>> >
>>
>> I've no problem with the patch anyway.
>
> Nor do I.
>
>> > - We seem to en
On Fri, 19 Feb 2010, Mel Gorman wrote:
> > > The patch below sets a smaller value for RECLAIM_DISTANCE and thus enables
> > > zone reclaim.
> >
>
> I've no problem with the patch anyway.
Nor do I.
> > - We seem to end up racing between zone_watermark_ok, zone_reclaim and
> > buffered_rmqueue.
On Fri, Feb 19, 2010 at 11:07:30AM +1100, Anton Blanchard wrote:
>
> Hi,
>
> > The patch below sets a smaller value for RECLAIM_DISTANCE and thus enables
> > zone reclaim.
>
I've no problem with the patch anyway.
> FYI even with this enabled I could trip it up pretty easily with a multi
> thre
Hi,
> The patch below sets a smaller value for RECLAIM_DISTANCE and thus enables
> zone reclaim.
FYI even with this enabled I could trip it up pretty easily with a multi
threaded application. I tried running stream across all threads in node 0. The
machine looks like:
node 0 cpus: 0 1 2 3 4 5 6
I noticed /proc/sys/vm/zone_reclaim_mode was 0 on a ppc64 NUMA box. It gets
enabled via this:
/*
* If another node is sufficiently far away then it is better
* to reclaim pages in a zone before going off node.
*/
if (distance > RECLAIM_DISTANCE)
14 matches
Mail list logo
