From owner-freebsd-hardware Mon Sep 4 12:43:54 2000 Delivered-To: freebsd-hardware@freebsd.org Received: from cage.simianscience.com (cage.simianscience.com [64.7.134.1]) by hub.freebsd.org (Postfix) with ESMTP id 0C38337B422; Mon, 4 Sep 2000 12:43:52 -0700 (PDT) Received: from chimp (chimp [192.168.0.2]) by cage.simianscience.com (8.11.0/8.9.3) with ESMTP id e84JhoV18765; Mon, 4 Sep 2000 15:43:50 -0400 (EDT) (envelope-from mike@sentex.net) Message-Id: <4.2.2.20000904152548.03405b38@mail.sentex.net> X-Sender: mdtancsa@mail.sentex.net X-Mailer: QUALCOMM Windows Eudora Pro Version 4.2.2 Date: Mon, 04 Sep 2000 15:38:34 -0400 To: Mike Smith From: Mike Tancsa Subject: Re: Problems installing 4.1 on 3ware RAID 5200 & 6200 Cc: "FreeBSDHW" In-Reply-To: <200009041934.MAA05065@mass.osd.bsdi.com> References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed Sender: owner-freebsd-hardware@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org At 12:34 PM 9/4/2000 -0700, Mike Smith wrote: > > I haven't tried RAID 0. The only other configuration I tried was installing > > to a single drive (not in an array, but still connected to the 3Ware > > controller). That worked just fine which is why I haven't shipped the card > > back yet.Yes, doing newfs on a 30 GB partition takes upwards of an hour. > >The newfs issue tends to be misleading; it's not representative of the >performance of the controller at all. > >If you newfs a redundant array that's background-initialising, it's slow. >(Slower than it really should be, IMO.) However the real killer is that >newfs uses the physio interface, and that runs into this issue from the >twe(4) manpage: > >BUGS >... > The controller cannot handle I/O transfers that are not aligned to a > 512-byte boundary. In order to support raw device access from user- > space, the driver will perform alignment fixup on non-aligned > data. This > process is inefficient, and thus in order to obtain best performance us- > er-space applications accessing the device should do so with aligned > buffers. > >Newfs doesn't use aligned buffers... Thanks for pointing this out. One thing that is not clear to me, how does this explain the very poor performance in bonnie ? If the poor bonnie results were being triggered by the alignment problem, would it not show equally poor results in RAID0 mode as well ? ---Mike -------------------------------------------------------------------- Mike Tancsa, tel +1 519 651 3400 Network Administration, mike@sentex.net Sentex Communications www.sentex.net Cambridge, Ontario Canada www.sentex.net/mike To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hardware" in the body of the message