From owner-freebsd-hackers Tue Nov 10 22:48:04 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id WAA08607 for freebsd-hackers-outgoing; Tue, 10 Nov 1998 22:48:04 -0800 (PST) (envelope-from owner-freebsd-hackers@FreeBSD.ORG) Received: from allegro.lemis.com (allegro.lemis.com [192.109.197.134]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id WAA08563 for ; Tue, 10 Nov 1998 22:47:58 -0800 (PST) (envelope-from grog@freebie.lemis.com) Received: from freebie.lemis.com (freebie.lemis.com [192.109.197.137]) by allegro.lemis.com (8.9.1/8.9.0) with ESMTP id QAA21144; Wed, 11 Nov 1998 16:20:11 +1030 (CST) Received: (from grog@localhost) by freebie.lemis.com (8.9.1/8.9.0) id QAA20721; Wed, 11 Nov 1998 16:20:00 +1030 (CST) Message-ID: <19981111162000.O20374@freebie.lemis.com> Date: Wed, 11 Nov 1998 16:20:00 +1030 From: Greg Lehey To: john cooper , hackers@FreeBSD.ORG, mike@smith.net.au, ticso@cicely.de Subject: Re: SCSI vs. DMA33.. References: <98Nov11.134648jst.21907@ns.isi.co.jp> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.91.1i In-Reply-To: <98Nov11.134648jst.21907@ns.isi.co.jp>; from john cooper on Wed, Nov 11, 1998 at 01:41:00PM +0900 WWW-Home-Page: http://www.lemis.com/~grog Organization: LEMIS, PO Box 460, Echunga SA 5153, Australia Phone: +61-8-8388-8286 Fax: +61-8-8388-8725 Mobile: +61-41-739-7062 Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Wednesday, 11 November 1998 at 13:41:00 +0900, john cooper wrote: > > Hi, > Just wondering if anyone has any _objective_ opinion on > the performance of say wide SCSI2 vs. DMA33 IDE drives [running > on contemporary motherboards]. Depends on what you mean by "objective". > The theoretical throughputs of 40MBs and 33MBs don't tell me a whole > lot. I know SCSI was the choice for performance in the past, > however I'm curious what others are seeing in actual usage these > days. SCSI is still the performance choice, but the field is closer now. I have five drives on my main machine: wd0: 1223MB (2504880 sectors), 2485 cyls, 16 heads, 63 S/T, 512 B/S wd2: 6197MB (12692736 sectors), 12592 cyls, 16 heads, 63 S/T, 512 B/S ide_pci: generic_dmainit 0170:1: warning, IDE controller timing not set wdc1: unit 1 (wd3): , DMA, 32-bit, multi-block-16 wd3: 8063MB (16514064 sectors), 16383 cyls, 16 heads, 63 S/T, 512 B/S ahc0: rev 0x03 int a irq 11 on pci0.9.0 ahc0: aic7870 Single Channel A, SCSI Id=7, 16/255 SCBs da0 at ahc0 bus 0 target 0 lun 0 da0: Fixed Direct Access SCSI2 device da0: 10.0MB/s transfers (10.0MHz, offset 15), Tagged Queueing Enabled da0: 2063MB (4226725 512 byte sectors: 64H 32S/T 2063C) da1 at ahc0 bus 0 target 2 lun 0 da1: Fixed Direct Access SCSI2 device da1: 3.300MB/s transfers, Tagged Queueing Enabled da1: 4096MB (8388608 512 byte sectors: 64H 32S/T 4096C) I'm running 3.0-CURRENT (post-RELEASE), and as you can see Ultra DMA is enabled on the IDE drives. Here's what I get transferring 32 MB from each raw device: $ dd if=/dev/rwd0c bs=32k count=1000 of=/dev/null 32768000 bytes transferred in 6.938876 secs (4722379 bytes/sec) $ dd if=/dev/rwd2c bs=32k count=1000 of=/dev/null 32768000 bytes transferred in 3.214075 secs (10195157 bytes/sec) $ dd if=/dev/rwd3c bs=32k count=1000 of=/dev/null 32768000 bytes transferred in 3.278695 secs (9994220 bytes/sec) $ dd if=/dev/rda0c bs=32k count=1000 of=/dev/null 32768000 bytes transferred in 5.734632 secs (5714055 bytes/sec) $ dd if=/dev/rsd1c bs=32k count=1000 of=/dev/null 32768000 bytes transferred in 8.893227 secs (3684602 bytes/sec) Looks good for the IDE drives, doesn't it? They say, though, that SCSI drives work better with multiple requests outstanding... $ dd if=/dev/rsd0c bs=32k count=1000 of=/dev/null & dd if=/dev/rsd1c bs=32k count=1000 of=/dev/null [3] 20705 32768000 bytes transferred in 9.940641 secs (3296367 bytes/sec) 32768000 bytes transferred in 12.121225 secs (2703357 bytes/sec) $ dd if=/dev/rsd0c bs=32k count=1000 of=/dev/null & dd if=/dev/rsd1c bs=32k count=1000 of=/dev/null 32768000 bytes transferred in 9.940080 secs (3296553 bytes/sec) 32768000 bytes transferred in 12.080951 secs (2712369 bytes/sec) Well, that doesn't look spectacular. What about the IDE drives (both on the same controller, wdc1): $ dd if=/dev/rwd2c bs=32k count=1000 of=/dev/null & dd if=/dev/rwd3c bs=32k count=1000 of=/dev/null 32768000 bytes transferred in 3.710713 secs (8830648 bytes/sec) 32768000 bytes transferred in 3.711320 secs (8829204 bytes/sec) $ dd if=/dev/rwd2c bs=32k count=1000 of=/dev/null & dd if=/dev/rwd3c bs=32k count=1000 of=/dev/null 32768000 bytes transferred in 3.737548 secs (8767245 bytes/sec) 32768000 bytes transferred in 3.729290 secs (8786659 bytes/sec) I must say, I'm surprised. This makes it look like there's more of a performance hit with concurrent requests on SCSI than on IDE. Let's look at the performance hits as a percentage (there's some guesswork which is which, of course, but only a little): alone 2 together % drop wd2 10.2 8.8 14 wd3 10.0 8.8 12 da0 5.7 3.3 42 da1 3.7 2.7 23 OK, the controller I have isn't the newest, but there's not exactly a lot of data crossing: even with the two disks transferring by themselves, they're transferring less data than a single DHEA drive. Can anybody else think of a reason for this? I have a 2940 Ultra ("ahc0: rev 0x00 int a irq 14 on pci0.17.0") in another machine. The disks aren't Ultra SCSI; is there any reason to think it will perform better? Greg -- See complete headers for address, home page and phone numbers finger grog@lemis.com for PGP public key To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message