Received: (from major@localhost)
	by minnie.cs.adfa.edu.au (8.9.1/8.9.1) id JAA17760
	for pups-liszt; Sat, 30 Jan 1999 09:08:05 +1100 (EST)
Received: from math.uwaterloo.ca (kcwellsc@math.uwaterloo.ca [129.97.140.144])
	by minnie.cs.adfa.edu.au (8.9.1/8.9.1) with ESMTP id JAA17754
	for <pups@minnie.cs.adfa.edu.au>; Sat, 30 Jan 1999 09:07:55 +1100 (EST)
Received: (from kcwellsc@localhost)
	by math.uwaterloo.ca (8.8.8/8.8.8) id RAA18899;
	Fri, 29 Jan 1999 17:07:36 -0500 (EST)
From: Ken Wellsch <kcwellsc@math.uwaterloo.ca>
Message-Id: <199901292207.RAA18899@math.uwaterloo.ca>
Subject: Re: low-end vaxen and unix
To: allisonp@world.std.com
Date: Fri, 29 Jan 1999 17:07:36 -0500 (EST)
Cc: pups@minnie.cs.adfa.edu.au
In-Reply-To: <Pine.SGI.3.95.990129160049.19225A-100000@world.std.com> from "allisonp@world.std.com" at Jan 29, 99 04:06:28 pm
Organization: University of Waterloo, Math Faculty Computing Facility (Alumni)
X-Mailer: ELM [version 2.4 PL25]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-pups@minnie.cs.adfa.edu.au
Precedence: bulk

On a KA410 (VAXstation 2000 etc.) the MFM controller and the NCR 5380
both do DMA to a shared 16Kb private memory buffer.  You then have to
pull your data out into the regular VAXen memory.  I believe the Lance
chip (ethernet) is the only DMA to main memory capable device.

-- Ken

> [...]  There are however issues in that the hard disk
> interface and the SCSI chip use the same DMA channel and it would cause
> some performance degrdation. [...]
> 
> Allison