From owner-freebsd-questions Mon Dec 20 13:41:53 1999 Delivered-To: freebsd-questions@freebsd.org Received: from rail.web.net (rail.web.net [192.139.37.202]) by hub.freebsd.org (Postfix) with ESMTP id 8444015033 for ; Mon, 20 Dec 1999 13:41:51 -0800 (PST) (envelope-from rob@rail.web.net) Received: by rail.web.net (Postfix, from userid 1001) id D6E071C02; Mon, 20 Dec 1999 16:39:58 -0500 (EST) Date: Mon, 20 Dec 1999 16:39:58 -0500 From: Rob Ellis To: freebsd-questions@FreeBSD.ORG Subject: Re: DMA errors Message-ID: <19991220163958.A38319@web.net> Mail-Followup-To: freebsd-questions@FreeBSD.ORG References: <19991220124638.B37447@web.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 1.0pre3i In-Reply-To: <19991220124638.B37447@web.net> Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > In the bios, there's a chipset setting for 'IDE Ultra DMA Mode', > which can be set to DISABLED or AUTO; if I set it to AUTO, > with DMA enabled in the kernel (wd flags), I get get regular > errors like this: > > > wd1: DMA failure, DMA status 5 > > wd2: DMA failure, DMA status 5 It looks like even with the bios 'IDE Ultra DMA Mode' set to DISABLE, performance still improves when I set the DMA flag in the kernel... I didn't think to check that. ;-) rawio shows 'sequential read' numbers about double with kernel DMA enabled, even with the bios setting off. So the bios setting just interferes with the kernel DMA stuff ? Or is it setting UDMA/66...?? Anyway, it seems ok off. ># rawio -v 1 /dev/vinum/rlocal # DMA on (b0ff) >Test ID K/sec /sec %User %Sys %Total >RR anon 2993.3 186 0.1 1.2 1.3 16384 >SR anon 30355.2 1853 1.5 25.1 26.5 16384 ># rawio -v 1 /dev/vinum/rlocal # DMA off (90ff) >Test ID K/sec /sec %User %Sys %Total >RR anon 2901.1 179 0.2 1.0 1.2 16384 >SR anon 14721.8 899 0.4 7.8 8.2 16384 - Rob To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message