From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S267502AbUHMVRI (ORCPT ); Fri, 13 Aug 2004 17:17:08 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S267503AbUHMVRH (ORCPT ); Fri, 13 Aug 2004 17:17:07 -0400 Received: from e2.ny.us.ibm.com ([32.97.182.102]:51587 "EHLO e2.ny.us.ibm.com") by vger.kernel.org with ESMTP id S267502AbUHMVRB (ORCPT ); Fri, 13 Aug 2004 17:17:01 -0400 Date: Fri, 13 Aug 2004 14:16:14 -0700 From: "Martin J. Bligh" To: Nick Piggin cc: Jesse Barnes , akpm@osdl.org, linux-kernel@vger.kernel.org, steiner@sgi.com Subject: Re: [PATCH] allocate page caches pages in round robin fasion Message-ID: <75260000.1092431774@flay> In-Reply-To: <411CFB04.603@yahoo.com.au> References: <200408121646.50740.jbarnes@engr.sgi.com> <200408130859.24637.jbarnes@engr.sgi.com> <89760000.1092414010@[10.10.2.4]> <200408130934.20913.jbarnes@engr.sgi.com> <92140000.1092415652@[10.10.2.4]> <411CFB04.603@yahoo.com.au> X-Mailer: Mulberry/2.1.2 (Linux/x86) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org >> Well, either we're: >> >> 1. Falling back and putting all our most recent accesses off-node. >> >> or. >> >> 2. Not falling back and only able to use one node's memory for any one >> (single threaded) app. >> >> Either situation is crap, though I'm not sure which turd we picked right >> now ... I'd have to look at the code again ;-) I thought it was 2, but >> I might be wrong. >> > > I'm looking at this now. We are doing 1 currently. In theory, yes. In practice, I have a feeling kswapd will keep us above the level of free memory where we'd fall back to another zone to allocate, won't it? > There are a couple of issues. The first is that you need to minimise > regressions for when working set size is bigger than the local node. Good point ... that is, indeed, a total bitch to fix. > I have a patch going now that just reclaims use-once file cache before > going off node. Seems to help a bit for basic things that just push > pagecache through the system. It definitely reduces remote allocations > by several orders of magnitude for those cases. Makes sense, but doesn't the same thing make sense on a global basis? I don't feel NUMA is anything magical here ... M.