From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Wed, 8 Nov 2006 02:21:41 -0800 From: Paul Jackson Subject: Re: [TAKE] memory page_alloc zonelist caching speedup Message-Id: <20061108022141.447abc92.pj@sgi.com> In-Reply-To: <20061010081429.15156.77206.sendpatchset@jackhammer.engr.sgi.com> References: <20061010081429.15156.77206.sendpatchset@jackhammer.engr.sgi.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux-mm@kvack.org Return-Path: To: Paul Jackson Cc: linux-mm@kvack.org, akpm@osdl.org, nickpiggin@yahoo.com.au, rientjes@google.com, ak@suse.de, mbligh@google.com, rohitseth@google.com, menage@google.com, clameter@sgi.com List-ID: As discussed in a separate lkml thread: Avoid allocating during interleave from almost full nodes the suggestion was made by Andrew and Christoph to consider replacing the wall clock based zapping of this zonelist cache with something based on the rate of vm paging activity, such as perhaps the counters. PGALLOC_* and PGSCAN_* (see further vmstat.h). But it will be a few weeks before I can get to doing this; I've got to do some other stuff first. I'm assuming that, as a practical matter, for the short term, either "time base" works, so that this is not an urgent change. Of course, if any lurkers want to jump in and do this sooner, have at it. -- I won't rest till it's the best ... Programmer, Linux Scalability Paul Jackson 1.925.600.0401 -- 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