From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail191.messagelabs.com (mail191.messagelabs.com [216.82.242.19]) by kanga.kvack.org (Postfix) with SMTP id D49A46B004A for ; Wed, 29 Sep 2010 11:14:21 -0400 (EDT) Date: Wed, 29 Sep 2010 09:52:50 -0500 (CDT) From: Christoph Lameter Subject: Re: zone state overhead In-Reply-To: <20100929144159.GC14204@csn.ul.ie> Message-ID: References: <20100928050801.GA29021@sli10-conroe.sh.intel.com> <20100928133059.GL8187@csn.ul.ie> <20100929100307.GA14204@csn.ul.ie> <20100929141730.GB14204@csn.ul.ie> <20100929144159.GC14204@csn.ul.ie> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-mm@kvack.org To: Mel Gorman Cc: David Rientjes , Shaohua Li , linux-mm@kvack.org List-ID: On Wed, 29 Sep 2010, Mel Gorman wrote: > > The threshold is stored in the hot part of the per cpu page structure. > > > > And the consequences of moving it? In terms of moving, it would probably > work out better to move percpu_drift_mark after the lowmem_reserve and > put the threshold after it so they're at least similarly hot across > CPUs. If you move it then the cache footprint of the vm stat functions (which need to access the threshold for each access!) will increase and the performance sink dramatically. I tried to avoid placing the threshold there when I developed that approach but it always caused a dramatic regression under heavy load. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org