Re: Commodore 8296GD

From: Michał Pleban <>
Date: Sun, 30 Sep 2012 09:48:30 +0200
Message-ID: <>

Rhialto wrote:

> Sounds much like what the DWW board is doing. I posted about that in
> this very mailing list. It has 8 banks of 1KB. It (most likely) used the
> offset into character cell memory, as output by the CRTC, as index into
> a 1 KB block of bitmap memory, combining it with the vertical offset
> into the character (0-7) (also as output by the CRTC) to select which bank.
> (Except that the DWW was designed before the CRTC was used, but I'm sure
> the same counters were present in the design anyway)

Yes. However, this board is even more simple - no PIA, no I/O chips
whatsoever, just some TTL glue logic. Maybe it uses some CRTC address
outputs as signal lines (like the CBM-II does) because otherwise I don't
know how it would even turn the graphics on.

> Yes, that is what I read elsewhere too. The "extra 32K" aka the
> "inaccessible 32K". Not that they're totally inaccessible: those 32 K
> are the second part of the non-expansion RAM (the other 64 KB of the 128
> KB are expansion RAM), and therefore would map from $8000-$FFFF if there
> were not ROM and I/O there.

Yes, this is probable. So far I looked at the initialization code at
$9000, and the first thing it does, it copies some routines to RAM at
$8800. So most probably the bitmap data is in the extra RAM, and the
graphics operations require banking out the ROMs - thus the routines
itself must be copied to that RAM. I still do not know how it banks them
out, though.

> And in fact you can access parts of it: $8xxx is normal screen
> (character cell) memory. $9xxx and $Axxxx can be easily jumpered to be
> RAM as well (if you don't have EPROMs installed in the sockets of the
> same address). And iirc there is also a 'jumper' option (but not so
> easily accessible) to connect a user port line to the NOROM line and
> make it possible to map out all ROMs, thereby uncovering even more of
> that RAM).

Well, there are ROMs at $9000 and $A000 of course, the HiRes routines
and BASIC extension :-)

> Given this, the interesting issue gets to be: Where is the bitmap, and
> how is it accessed by the CPU?

Yes, that's what I'm trying to figure out :-)

> Given the resolution you mention (512 x 256) it would need 16 KB of
> memory, or half of that 32 KB space. If normal text output would remain
> working, and perhaps they thought it was nice if the $9xxx and $Axxx
> would keep working as it was, it should be at $C000-$FFFF. Somehow.

My guess too. And yes, the normal screen still works independently of
the graphics - if you turn out the graphics, the text screen remains as
it was (and vice versa).

> I wonder why that resolution was chosen, and not 640 x 200 (which is 8 x
> (80 x 25)). That would make it much more compatible with the text
> screen.

Maybe to retain compatibility with boards based on the Thomson chip,
which used that resolution?


       Message was sent through the cbm-hackers mailing list
Received on 2012-09-30 08:00:04

Archive generated by hypermail 2.2.0.