From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Wed, 2 Jan 2008 22:44:50 -0500 From: Rik van Riel Subject: Re: [patch 00/19] VM pageout scalability improvements Message-ID: <20080102224450.585bf956@bree.surriel.com> In-Reply-To: <20080102224144.885671949@redhat.com> References: <20080102224144.885671949@redhat.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: linux-kernel@vger.kernelporg Cc: linux-mm@kvack.org, lee.schermerhorn@hp.com List-ID: On Wed, 02 Jan 2008 17:41:44 -0500 linux-kernel@vger.kernelporg wrote: > This patch series improves VM scalability by: > 3) switching to SEQ replacement for the anonymous LRUs, so the > number of pages that need to be scanned when the system > starts swapping is bound to a reasonable number I have done some (minimal) testing of this tonight. Running a 16000 MB fillmem on my 16GB test box (where slub eats up unexplainable amounts of memory so the test gets about 14GB RSS and 1.5GB in swap). 2.6.24-rc6-mm1 $ time ./fillmem 16000 real 3m43.601s user 0m11.090s sys 0m55.274s 2.6.24-rc6-mm1-vmsplit $ time ./fillmem 16000 real 1m51.323s user 0m10.638s sys 0m42.859s This is after carefully emptying out all memory by rebooting the system and running fillmem a few times. These results are repeatable. With vanilla 2.6.24-rc6-mm1, kswapd eats up 99% of the CPU at a few points in the test run. This never seems to happen with the vmsplit kernel. Unfortunately the symbol file for vanilla 2.6.24-rc6-mm1 seems to be hosed, so readprofile is not returning anything useful on that kernel. I'll try to come up with more useful data later. -- All rights reversed. -- 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