[Coco] Christmas Wish Device

Willard Goosey goosey at virgo.sdc.org
Thu Dec 8 03:58:03 EST 2011


On Wed, Dec 07, 2011 at 06:56:13PM -0600, Robert Hermanek wrote:
> I don't think my thought comes across too clearly, maybe because it is 
> impossible for some reasons I don't understand.  But I'm not thinking of 
> the cart RAM as something the coco could access directly, as in available 
> to machine language instructions like "real" RAM  --instead it would be a 
> device that needs to be controlled through i/o registers like a stereo 
> pack, rs232 or anything else.  So in my example:

Sure, such a device could be built.  

I think a lot of the thrashing around in this thread is people trying to
figure out why you'd want such a thing.  I'm kind of confused myself.

[..]

> The only downside I can see is you would have to use 5 precious register 
> addresses in this example... 

The downside you're not seeing is the speed penalty for such a device.
You say you don't want to go through the block-oriented hoops of a
disk controller, but the reason those hoops exist is to make transfers
as fast as possible.  Updating your "address register" for each byte
is going to be very slow. (Either that or you have an intelligent MMU
that does auto increment/deincrement and you have to program it and
suddenly it's as complicated [and not nearly as well-debugged] as the
disk controller you were bashing.)

Yes, I have thought about this a fair bit.  I had an idea for a
hardware project where a block of memory like this would be used.  I
finally concluded it wouldn't be much use.

>anyway, don't mean to go on and on about 
> something I couldn't build anyway, just dreaming out loud.

Willard
-- 
Willard Goosey  goosey at sdc.org
Socorro, New Mexico, USA
I search my heart and find Cimmeria, land of Darkness and the Night.
  -- R.E. Howard



More information about the Coco mailing list