From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755847Ab1FTTmk (ORCPT ); Mon, 20 Jun 2011 15:42:40 -0400 Received: from one.firstfloor.org ([213.235.205.2]:42503 "EHLO one.firstfloor.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755323Ab1FTTmi (ORCPT ); Mon, 20 Jun 2011 15:42:38 -0400 Date: Mon, 20 Jun 2011 21:42:36 +0200 From: Andi Kleen To: Christoph Lameter Cc: Pekka Enberg , David Rientjes , Andi Kleen , tj@kernel.org, Metathronius Galabant , Matt Mackall , Eric Dumazet , Adrian Drzewiecki , linux-kernel@vger.kernel.org Subject: Re: [slub p2 0/4] SLUB: [RFC] Per cpu partial lists V2 Message-ID: <20110620194236.GA32765@one.firstfloor.org> References: <20110620153244.214038140@linux.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20110620153244.214038140@linux.com> User-Agent: Mutt/1.4.2.2i Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Jun 20, 2011 at 10:32:44AM -0500, Christoph Lameter wrote: > The following patchset applied on top of the lockless patchset V7. It > introduces per cpu partial lists which allow a performance increase of > around ~15 during contention for the nodelock (can be tested using > hackbench). What size system did you test it on? > > These lists help to avoid per nodelocking overhead. Allocator latency > could be further reduced by making these operations work without > disabling interrupts (like the fastpath and the free slowpath) as well as > implementing better ways of handling ther cpu array with partial pages. I think we really need better batching for the transfers. -andi -- ak@linux.intel.com -- Speaking for myself only.