From owner-freebsd-scsi Wed May 20 10:18:09 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id KAA05538 for freebsd-scsi-outgoing; Wed, 20 May 1998 10:18:09 -0700 (PDT) (envelope-from owner-freebsd-scsi@FreeBSD.ORG) Received: from sendero.simon-shapiro.org (sendero.simon-shapiro.org.142.69.207.in-addr.arpa [207.69.142.25] (may be forged)) by hub.freebsd.org (8.8.8/8.8.8) with SMTP id KAA05465 for ; Wed, 20 May 1998 10:17:57 -0700 (PDT) (envelope-from shimon@sendero.simon-shapiro.org) Received: (qmail 26199 invoked by uid 1000); 20 May 1998 18:19:06 -0000 Message-ID: X-Mailer: XFMail 1.3-beta-042198 [p0] on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: Date: Wed, 20 May 1998 14:19:06 -0400 (EDT) Reply-To: shimon@simon-shapiro.org Organization: The Simon Shapiro Foundation From: Simon Shapiro To: Raul Zighelboim Subject: RE: DPT driver performance Cc: "n@nectar.com" , "scsi@freebsd.org" , Tom Sender: owner-freebsd-scsi@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On 30-Apr-98 Raul Zighelboim wrote: > > I am running ONE drive on an AHC controller and ONE drive on a DPT > controller. No stripping, no raid. Just one drive. Both are formated > with 512b/s. > > No, I do not intend to run iozone as my main application. But it > bothers me that under the same conditions, the Adaptec controller can > write a large file twice as fast. > > You're saying that this is not relevant ? What you see is normal. Consider that the DPT has its own CPU, O/S, and SCSI controller chip. There are certain operations in which the latencies (as minute as they may be) introduced here are visible. The AHC controllers are directly attached to the system bus. If you intend to run only a single drive, on a desktop, I would say that a DPT is a costly and ineffective solution. If your needs are for RAID arrays, large disk farms, very heavy concurrent load, the DPT will serve you rather well. When testing/certifying a new driver release, I never use any of the standard benchmarks, for a reason; They are mostly geared towards sequential access to large files. Most of the DPTs are used in a heavily random seek environment or in a ``cannot fail'' environment. For these, I use my own mad testing procedure, which has been posted here more than once. I apologize for the late reply, but today is the second day since my move and the first in which I read email. Simon To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message