From owner-freebsd-current Thu May 29 08:53:34 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id IAA21827 for current-outgoing; Thu, 29 May 1997 08:53:34 -0700 (PDT) Received: from wlk.com (news.wlk.com [192.86.83.250]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id IAA21821 for ; Thu, 29 May 1997 08:53:31 -0700 (PDT) Received: from SMTPdaemon by wlk.com (smail3.2) with SMTPL id m0wX7Vu-0009sVC; Thu, 29 May 1997 10:53:18 -0500 (CDT) Received: from critter.dk.tfs.com (localhost [127.0.0.1]) by critter.dk.tfs.com (8.8.5/8.8.5) with ESMTP id RAA00375; Thu, 29 May 1997 17:06:23 +0200 (CEST) To: Bruce Evans cc: current@freebsd.org From: Poul-Henning Kamp Subject: Re: disk cache challenged by small block sizes In-reply-to: Your message of "Thu, 29 May 1997 22:28:05 +1000." <199705291228.WAA11504@godzilla.zeta.org.au> Date: Thu, 29 May 1997 17:06:22 +0200 Message-ID: <373.864918382@critter.dk.tfs.com> Sender: owner-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk >ufs with a block size of 4K was about the same speed as ext2fs with a block >size of 4K. ufs with a block size of 8K was significantly (25%) slower. =========================================================== Uhm, isn't that rather obvious ? The chances of being able to do one DMA into two (physically) consecutive pages are very very slim as far as I can tell, so you will generally get better performance when you do it in page size chunks ? In other words, drivers should be optimized for two back-to-back 4K transactions to different pages, not for 8k transactions to physically contiguous pages. -- Poul-Henning Kamp | phk@FreeBSD.ORG FreeBSD Core-team. http://www.freebsd.org/~phk | phk@login.dknet.dk Private mailbox. whois: [PHK] | phk@tfs.com TRW Financial Systems, Inc. Power and ignorance is a disgusting cocktail.