From owner-freebsd-hackers Wed Mar 22 09:32:48 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.10/8.6.6) id JAA10317 for hackers-outgoing; Wed, 22 Mar 1995 09:32:48 -0800 Received: from ref.tfs.com (ref.tfs.com [140.145.254.251]) by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id JAA10311 for ; Wed, 22 Mar 1995 09:32:47 -0800 Received: (from phk@localhost) by ref.tfs.com (8.6.8/8.6.6) id JAA05890; Wed, 22 Mar 1995 09:32:31 -0800 From: Poul-Henning Kamp Message-Id: <199503221732.JAA05890@ref.tfs.com> Subject: Re: Why IDE is bad To: bde@zeta.org.au (Bruce Evans) Date: Wed, 22 Mar 1995 09:32:30 -0800 (PST) Cc: freebsd-hackers@freefall.cdrom.com, kuku@gilberto.physik.rwth-aachen.de In-Reply-To: <199503221132.VAA11806@godzilla.zeta.org.au> from "Bruce Evans" at Mar 22, 95 09:32:15 pm Content-Type: text Content-Length: 814 Sender: hackers-owner@FreeBSD.org Precedence: bulk > >Regarding the difference in CPU utilization, could it be - I don't > >know the driver internals - that it's the bounce buffer technique > >that costs CPU while the SCSI controller uses bus master transfers all > >the time? Or were you comparing VLB EIDE vs. SCSI ? > > This might explain why my Intr time is so much lower than Poul's. > I have only 16MB, and don't use option BOUNCE_BUFFERS. The bcopy() > for bouncing is done in a call from biodone(). biodone() is called > from the interrupt handler, at least for the wd driver. Excuse moi! Why would we need bounce-buffers for wd.c when the IO is done using "REP INSW" ?? -- Poul-Henning Kamp -- TRW Financial Systems, Inc. 'All relevant people are pertinent' && 'All rude people are impertinent' => 'no rude people are relevant'