Path: utzoo!attcan!uunet!tut.cis.ohio-state.edu!cs.utexas.edu!rutgers!cbmvax!daveh
From: daveh@cbmvax.commodore.com (Dave Haynie)
Newsgroups: comp.sys.amiga.hardware
Subject: Re:  Amiga 4000
Message-ID: <13019@cbmvax.commodore.com>
Date: 3 Jul 90 18:47:11 GMT
References: <2386@speedy.mcnc.org>
Reply-To: daveh@cbmvax (Dave Haynie)
Distribution: na
Organization: Commodore, West Chester, PA
Lines: 46

In article <2386@speedy.mcnc.org> rlsmith@mcnc.org (Robert L. Smith) writes:

>    The other major improvement that all Amigas need is a hardware
>change enabling use of the principal 68K-family program debugging
>tool.  I'm speaking of Bus Error, returned for an unrecognized value
>on the Address Bus.  

While that's hardly the principal 68K family program debugging tool (most
debugging tools are far more sophisticated, and on any 68030 machine, the
generation of the level 2 exception that this bus error circuitry gives
you can just as easily be added via software traps).

>A single-shot plus a trifle of logic would implement that, especially if 
>DTACK were properly left to each peripheral.

Well, of course there's no DTACK* on the 3000 execept on the expansion
bus.  But we do have several bus timeout options offered via a programmable
register in the GARY chip.  Basically, at powerup, any cycle that isn't
terminated by DSACKn*, STERM*, AVEC*, or any externally generated BERR*
will get both DSACKs yanked for it after an 8uS timeout.  This mode allows
software to poll various memory areas that might actually contain things,
without wasting an undue amount of time during boot up.  Once the system
is up, the timeout mode is changed to generate a BERR* after 1/4 Second with
no proper cycle termination.

>    When Commodore wants to get serious, it should seriously consider
>these fundamentals.

Well, I don't want to restart YALDOPD (Yet Another Long Drawn-Out Parity
Discussion).  But regardless of it's actual utility, most of the folks who
think they need parity (like the bankers you mention) don't have a single
clue about the actual tradeoffs involved.  They are simply superstitious.
Which makes the issue of parity a Marketing issue, something I'm not going
to speak on.  If Marketing wants it, Marketing will have it.  

Of course, any expansion memory board can support parity memory, and fail
just as miserably as any PClone in the extremely unlikely event it encounters
a parity failure.

>Regards, rLs


-- 
Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: hazy     BIX: hazy
	"I have been given the freedom to do as I see fit" -REM