From owner-freebsd-hackers Tue Aug 27 09:18:39 1996 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id JAA12587 for hackers-outgoing; Tue, 27 Aug 1996 09:18:39 -0700 (PDT) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.211]) by freefall.freebsd.org (8.7.5/8.7.3) with SMTP id JAA12580 for ; Tue, 27 Aug 1996 09:18:35 -0700 (PDT) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id JAA24772; Tue, 27 Aug 1996 09:02:50 -0700 From: Terry Lambert Message-Id: <199608271602.JAA24772@phaeton.artisoft.com> Subject: Re: Multiple swaps slow down system? To: koshy@india.hp.com (Koshy) Date: Tue, 27 Aug 1996 09:02:49 -0700 (MST) Cc: hal@post.vale.com, freebsd-hackers@freebsd.org In-Reply-To: <199608270327.AA078696441@cyborg.india.hp.com> from "Koshy" at Aug 27, 96 08:27:18 am X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-hackers@freebsd.org X-Loop: FreeBSD.org Precedence: bulk > IDE (ATA-2) can use DMA, in fact, future ATA standards make DMA support > mandatory. I was under the impression that probing the chipset for this capability will crash older chipsets. Thus it is impossible to probe for the capability uniformly and safely, other than by keeping a ROM/ID table and using that to detect the things. Windows95 and NT get around the issue by logging probes during install. If a user resets after an "if this takes too long..." message, it decides the hardware isn't there when the install starts up again (what the hell is "too long"? ...they don't tell us). Further, the RZ1000 and two other IDE chipsets are well known to silently fail DMA's if an interrupt occurs during a DMA transfer. DOS does not have this problem, and Windows95 and NT 4.x beta both silently turn the feature off in the user's CMOS. There is "test" software on the Intel www site -- it *also* silently turns off the feature where it can, claiming "you don't have to worry about the problem". A careful examination of the CMOS settings before and afer will show that the twiddling has taken place (AMI BIOS has an advanced setup screen for IDE-on-the-motherboard BIOS for turning on or of "overlapped I/O" -- this seeting has been seen to change). > However, IDE devices sell for around half the price of comparable > SCSI devices in most parts of the world, so they are probably going > to be around for a while. I always wonder if this is just dealers milking the market, or what. The latest "Processor" magazine shows no price difference in drives over 250M or so. It is just impossible to subscribe to "Processor" or "Computer Shopper" outside the US, or just impossible to buy through them? Or is it possible, but just not done, in an international conspiracy to portray SCSI as costing more? 8-). Regards, Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.