From owner-freebsd-stable Fri May 25 11: 1:28 2001 Delivered-To: freebsd-stable@freebsd.org Received: from solfertje.student.utwente.nl (wit401310.student.utwente.nl [130.89.236.150]) by hub.freebsd.org (Postfix) with ESMTP id B86F037B616 for ; Fri, 25 May 2001 11:01:20 -0700 (PDT) (envelope-from dalroi@wit401310.student.utwente.nl) Received: from solfertje.student.utwente.nl (localhost.student.utwente.nl [127.0.0.1]) by solfertje.student.utwente.nl (Postfix) with ESMTP id E885A1F63; Fri, 25 May 2001 20:02:30 +0200 (CEST) Date: Fri, 25 May 2001 20:02:28 +0200 (CEST) From: Alban Hertroys Subject: Re: Athlon Power and FreeBSD-STABLE experiences. To: Salvo Bartolotta Cc: freebsd-stable@freebsd.org, Ron Klinkien In-Reply-To: <20010525.16051800@bartequi.ottodomain.org> MIME-Version: 1.0 Content-Type: TEXT/plain; CHARSET=US-ASCII Message-Id: <20010525180230.E885A1F63@solfertje.student.utwente.nl> Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG On 25 May, Salvo Bartolotta wrote: >>>>>>>>>>>>>>>>>>> Original Message <<<<<<<<<<<<<<<<<< > > Is anyone currently running 4-S on this hardware, with **IDE** disks? Yes. It runs OK if you turn off "Delayed transactions". You will probably get data-corruption on your HDD's if you leave that setting on. That is supposedly a problem of the VIA 686B south-bridge, which is used on most AMD760 based boards (including the ASUS A7M266). It's used on almost every recent socket A board. The problem is amplified if you have a Creative Labs SBLive, which I have (a Live 1024). My setup consists of only one IDE HDD (IBM DTLA 46GB, 7200rpm), and two ATAPI-devices, which I hardly use in FreeBSD (just for games/DVD's in Windows), so there's not much data traffic between the two IDE controllers in the testing environment. At first I thought I had a memory problem, so I wrote a small test-program in C to test my system memory. That program turned out to be quite good at testing this problem (IMO). It writes test patterns in random order in a large amount of allocated space, and then tries to read it back all at once. If you need swapspace (~30MB) while testing, the machine will panic, fail to sync it's discs and reboot - which is definately not good. After turning off "Delayed transactions", the machine was stable. (The memory tests succeeded, BTW). I can make the source available if wanted. Funny thing is that I didn't have any trouble in Windows98. I don't know what that means, but I have my suspicions ;) My source of information on the problem was this article: http://www.theregister.co.uk/content/3/18267.html The article itself doesn't reveal too much about the problem and is mostly biased to possible problems when running Windows, but the german articles it links to are worth the read (if you understand german). Does anybody know more? (Like: Is there still data-corruption without delayed transactions? What is the benefit of delayed transactions in a 'PCI-only' system? Why didn't anybody use the AMD southbridge? Is there intelligent life out there?) Sorry if the lengthy answer is inconvenient to someone. I'm in a writing mood, and I don't consider that harmful to anybody but myself. -- Alban Hertroys http://solfertje.student.utwente.nl - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Install a BSD and become religious. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message