From owner-freebsd-hardware@FreeBSD.ORG Tue Apr 17 23:01:36 2012 Return-Path: Delivered-To: freebsd-hardware@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id E233C106564A for ; Tue, 17 Apr 2012 23:01:36 +0000 (UTC) (envelope-from dieterbsd@engineer.com) Received: from mailout-us.gmx.com (mailout-us.gmx.com [74.208.5.67]) by mx1.freebsd.org (Postfix) with SMTP id 87FF98FC08 for ; Tue, 17 Apr 2012 23:01:36 +0000 (UTC) Received: (qmail 23228 invoked by uid 0); 17 Apr 2012 23:01:30 -0000 Received: from 67.206.184.108 by rms-us014.v300.gmx.net with HTTP Content-Type: text/plain; charset="utf-8" Date: Tue, 17 Apr 2012 19:01:28 -0400 From: "Dieter BSD" Message-ID: <20120417230129.155080@gmx.com> MIME-Version: 1.0 To: freebsd-hardware@freebsd.org X-Authenticated: #74169980 X-Flags: 0001 X-Mailer: GMX.com Web Mailer x-registered: 0 Content-Transfer-Encoding: 8bit X-GMX-UID: HpP/b/Bd3zOlNR3dAHAhFQF+IGRvbwA3 Subject: Re: PCI-X SATA (non HW-RAID) controller recommendation X-BeenThere: freebsd-hardware@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: General discussion of FreeBSD hardware List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 17 Apr 2012 23:01:37 -0000 > having SMART work is probably a good idea Note that some (most?) of the USB-to-SATA bridges do not provide access to SMART, at least not on FreeBSD. Also can't turn off the write buffer, and no NCQ, and pathetically slow. >> I have the Silicon Image 3132 which is PCIe-x1 with 2 sata ports. >> Not as fast as it should be but fast enough for my needs. >> Works well with FreeBSD siis(4), which provides NCQ. >> Works well with the 3726 port multiplier. Talks to recent >> 600MB/s drives at 300MB/s, unlike JMB363 which doesn't like >> 600MB/s drives, even with the sata rev hint set. > > 300MB/s should be enough. With vanilla rotating drives even 150 is enough. Speed-wise, NCQ is far more important than 300 or 600. Problem is that recent drives are 600 and don't work with JMB363. Drives used to have a jumper to pretend to only talk 150, for controllers that didn't deal well with 300. But as far as I know there isn't any way to get the 600 drives to pretend to be only 150 or 300. >> Issue: if a port has a problem (flaky disk or whatever), siis(4) may >> do a bunch of DELAY(big number) which interferes with other hardware >> doing real-time data logging, causing data to be lost. Unacceptable. >> Does not require power cycle though. I don't recall it even needing >> a reboot. > > That doesn't sound good, but I guess somebody can tell me the issues of > any controller or driver that comes up in this discussion. If you're OK > with siis despite this, I should too. It isn't acceptable, but I've had similar/identical problems with ata(4) and ahci(4). For all I know maybe all the disk drivers do it. I think the problem is keeping interrupts off for too long.