Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84 / ST 1.0; site saber.UUCP
Path: utzoo!watmath!clyde!bonnie!akgua!whuxlm!harpo!decvax!decwrl!sun!idi!saber!skinner
From: skinner@saber.UUCP (Robert Skinner)
Newsgroups: net.micro.cbm,net.micro,net.rumor
Subject: Re: Flames Re: "Valid" AMIGA info.
Message-ID: <1644@saber.UUCP>
Date: Thu, 20-Jun-85 13:27:18 EDT
Article-I.D.: saber.1644
Posted: Thu Jun 20 13:27:18 1985
Date-Received: Sun, 23-Jun-85 02:50:56 EDT
References: <1655@amdahl.UUCP> <5706@utzoo.UUCP>
Organization: Saber Technology, San Jose, CA
Lines: 41
Xref: watmath net.micro.cbm:1484 net.micro:10844 net.rumor:928

> ***FLAME WARNING***  Those who are easily singed should read no further.
> 
> >     One of the nicest features of  the
> > graphics hardware is that it uses  the
> > "off-phase"  periods  of  the   system
> > clock, when the processor is not using
> > the  bus.
> 
> This is really interesting, since the 68000 *has* no "off phase", unlike
> the Motorola 8-bit chips.  In fact, the 68000 is pretty much bus-bound,
> and there isn't a lot of spare bandwidth to go around.  There is a smell
> of snake oil in the air...
>
You are right that the 68000 has no explicit "off phase", but I wouldn't
call it bus-bound either.  Just because the address lines are active, the 
processor doesn't have to hog memory.

As an example, a couple of companies ago I helped build a graphics terminal
with the 68000 and 640x480 resolution.  60Hz refresh required a pixel rate
of about 28MHz, which implied that we did a 16-bit video read every 570ns.
Now, 570ns is a long time for DRAMS, so we double-buffered the memory, with
280ns (3.5MHz) access cycles and let the video access memory at regular 
intervals with priority.  We put the 68000 clock at 7MHz and it syncs
up to the memory access slot allotted for it.  It's true that the 68000
asserted it's memory lines about half way through the video memory cycle,
but the arbitration logic didn't switch them in until the appropriate time.
So, the two systems co-existed and the 68000 was forced to wait only about
%5 of the time, because very few instructions take a non-multiple of 4 cycles.

I know that 7MHz is not blazing for a vanilla 68K, but 640x480 resolution is
good for a pc.  Of course there are trade offs that can be made:  the clock
speed could be set at 14MHz and video could fetch 32-bit words.  The point is
that 68K systems work with double-buffered memory without *terrible* penalties,
and no snake oil either.

-----------------------------------------------------------------------------
Name:	Robert Skinner
Mail:	Saber Technology, 2381 Bering Drive, San Jose, California 95131
AT&T:	(408) 945-0518, or 945-9600 (mesg. only)
UUCP:	...{decvax,ucbvax}!decwrl!saber!skinner
	...{amd,ihnp4,ittvax}!saber!skinner