Re: [PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-07-01 Thread Mel Gorman
On Mon, Jun 30, 2014 at 03:09:14PM -0700, Andrew Morton wrote: > On Mon, 30 Jun 2014 22:51:21 +0100 Mel Gorman wrote: > > > > That's a large change in system time. Does this all include kswapd > > > activity? > > > > > > > I don't have a profile to quantify that exactly. It takes 7 hours to >

Re: [PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-07-01 Thread Mel Gorman
On Mon, Jun 30, 2014 at 03:09:14PM -0700, Andrew Morton wrote: On Mon, 30 Jun 2014 22:51:21 +0100 Mel Gorman mgor...@suse.de wrote: That's a large change in system time. Does this all include kswapd activity? I don't have a profile to quantify that exactly. It takes 7 hours to

Re: [PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-06-30 Thread Andrew Morton
On Mon, 30 Jun 2014 22:51:21 +0100 Mel Gorman wrote: > > That's a large change in system time. Does this all include kswapd > > activity? > > > > I don't have a profile to quantify that exactly. It takes 7 hours to > complete a test on that machine in this configuration That's nuts. Why

Re: [PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-06-30 Thread Mel Gorman
On Mon, Jun 30, 2014 at 02:14:04PM -0700, Andrew Morton wrote: > On Mon, 30 Jun 2014 17:48:03 +0100 Mel Gorman wrote: > > > The fair zone allocation policy round-robins allocations between zones > > within a node to avoid age inversion problems during reclaim. If the > > first allocation fails,

Re: [PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-06-30 Thread Andrew Morton
On Mon, 30 Jun 2014 17:48:03 +0100 Mel Gorman wrote: > The fair zone allocation policy round-robins allocations between zones > within a node to avoid age inversion problems during reclaim. If the > first allocation fails, the batch counts is reset and a second attempt > made before entering the

[PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-06-30 Thread Mel Gorman
The fair zone allocation policy round-robins allocations between zones within a node to avoid age inversion problems during reclaim. If the first allocation fails, the batch counts is reset and a second attempt made before entering the slow path. One assumption made with this scheme is that

[PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-06-30 Thread Mel Gorman
The fair zone allocation policy round-robins allocations between zones within a node to avoid age inversion problems during reclaim. If the first allocation fails, the batch counts is reset and a second attempt made before entering the slow path. One assumption made with this scheme is that

Re: [PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-06-30 Thread Andrew Morton
On Mon, 30 Jun 2014 17:48:03 +0100 Mel Gorman mgor...@suse.de wrote: The fair zone allocation policy round-robins allocations between zones within a node to avoid age inversion problems during reclaim. If the first allocation fails, the batch counts is reset and a second attempt made before

Re: [PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-06-30 Thread Mel Gorman
On Mon, Jun 30, 2014 at 02:14:04PM -0700, Andrew Morton wrote: On Mon, 30 Jun 2014 17:48:03 +0100 Mel Gorman mgor...@suse.de wrote: The fair zone allocation policy round-robins allocations between zones within a node to avoid age inversion problems during reclaim. If the first allocation

Re: [PATCH 4/4] mm: page_alloc: Reduce cost of the fair zone allocation policy

2014-06-30 Thread Andrew Morton
On Mon, 30 Jun 2014 22:51:21 +0100 Mel Gorman mgor...@suse.de wrote: That's a large change in system time. Does this all include kswapd activity? I don't have a profile to quantify that exactly. It takes 7 hours to complete a test on that machine in this configuration That's nuts.