From owner-freebsd-scsi@FreeBSD.ORG Tue May 11 14:01:31 2004 Return-Path: Delivered-To: freebsd-scsi@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id AB45A16A4CE for ; Tue, 11 May 2004 14:01:31 -0700 (PDT) Received: from pooker.samsco.org (pooker.samsco.org [168.103.85.57]) by mx1.FreeBSD.org (Postfix) with ESMTP id A4A2743D49 for ; Tue, 11 May 2004 14:01:30 -0700 (PDT) (envelope-from scottl@freebsd.org) Received: from freebsd.org (junior-wifi.samsco.home [192.168.0.11]) by pooker.samsco.org (8.12.10/8.12.10) with ESMTP id i4BL6Hu6062841; Tue, 11 May 2004 15:06:18 -0600 (MDT) (envelope-from scottl@freebsd.org) Message-ID: <40A13F22.80505@freebsd.org> Date: Tue, 11 May 2004 15:01:22 -0600 From: Scott Long User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.6) Gecko/20040304 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Jens Schweikhardt References: <20040511203730.GB2256@schweikhardt.net> In-Reply-To: <20040511203730.GB2256@schweikhardt.net> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Spam-Status: No, hits=0.0 required=3.8 tests=none autolearn=no version=2.63 X-Spam-Checker-Version: SpamAssassin 2.63 (2004-01-11) on pooker.samsco.org cc: freebsd-scsi@freebsd.org Subject: Re: 2120S Stripe - abysmal performance X-BeenThere: freebsd-scsi@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: SCSI subsystem List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 11 May 2004 21:01:31 -0000 Jens Schweikhardt wrote: > hello, world\n > > I hoped that my new Adaptec 2120S U320 RAID in my new Supermicro P4SCT > board with 64bit/66MHz PCI would turn my disk IO into a desktop blitz, > but the opposite has happened. While my old ASUS A7N8X with vinum would > read 100MB/s from 2 Fujitsu MAP3367 connected to an 29160, I get only > 29MB/sec from the stripe built with the 2120S (as measured with dd to > /dev/null). > > What I have done so far: > o have reset the 2120S config to factory defaults. > o set the transfer speed to U160 and disabled "packetized" because > I could not low level format the drives otherwise. Until that point > I had a lot of trouble even getting the 2120S to boot its controller > kernel. Effect has been seemingly endless (5min) twirling baton. This means that you have a bad cable and/or poor termination. Backing down to u160 is only going to hurt performance; the 2120 needs every little bit of help that it can get. > o low level formatted both MAPs, initialized them and then built the > stripe with 64kb stripe size. > o I have enabled caching in both directions. Read caching should be turned off unless you have a very specific need for it. Having it on is going to hurt sequential read performance by a 2x factor. The only time is makes sense is when you have a relatively small data set that gets read repeatedly, with few other reads mixed in. > > The dmesg says > schweikh@hal9000:~ $ dmesg|grep aac > aac0: mem 0xe8000000-0xebffffff irq 9 at device 1.0 on pci3 > aac0: Reserved 0x4000000 bytes for rid 0x10 type 3 at 0xe8000000 > aac0: [FAST] > aac0: i960RX 100MHz, 48MB cache memory, optional battery present > aac0: Kernel 4.0-0, Build 6011, S/N ba12c0 > aac0: Supported Options=1f7e > aacd0: on aac0 > aacd0: 69997MB (143354880 sectors) > > Any enlightenment on how to improve disk io greatly appreciated. > The 2120 is the 'value' card of the family. It has an Intel 80302 instead of an 80303 processor, which means that it's internal i/o thoroughput is cut in half. Add in the 2x hit from having the read cache on, and yes, you will get horribly performance. At the very least you need to fix your cable configuration and turn off the read cache. Scott