A 65c816 as a co-processor

Alex Freed

unread,

Sep 25, 2012, 2:40:53 PM9/25/12

to

Looking at the latest Jameco catalog I noticed that they started
carrying the *new* 65C816 and 65C02 CPUs rated at 16 MHz! This is more
MIPS than a 20 MHz Z80 that could be used in my Applicard clone.

It would be relatively easy to make a card for A2e with this fast CPU
and 128K (if not 512K) of SRAM. No bank switching needed if 65C816 is
used. Talk to the main CPU via a mailbox, like the Applicard.

The only question is: does anyone need such a card? Given the ready
supply of Apple 2 GS computers?

-Alex.

D Finnigan

unread,

Sep 25, 2012, 3:04:17 PM9/25/12

to

Alex Freed wrote:
>
> It would be relatively easy to make a card for A2e with this fast CPU

...

> The only question is: does anyone need such a card? Given the ready
> supply of Apple 2 GS computers?
>

I would welcome the increased address space on an Apple IIe with the 65C816.

magnusfalkirk

unread,

Sep 25, 2012, 3:07:09 PM9/25/12

to

If you could set it up as an accelerator for the GS I'd be interested. I've got the stock 2.8MHz cpu in my GS.

Dean

Egan Ford

unread,

Sep 25, 2012, 3:19:39 PM9/25/12

to

On 9/25/12 12:40 PM, Alex Freed wrote:
> The only question is: does anyone need such a card?

Want, not need. Sign me up for one. Since using a mailbox I am
assuming that you'll provide some type of boot disk or boot loader.

Any chance of a shared memory mode like the old Z80s and the Mill 6809?

Thanks.

aiia...@gmail.com

unread,

Sep 26, 2012, 12:09:40 PM9/26/12

to

On Tuesday, September 25, 2012 11:40:47 AM UTC-7, Alex Freed wrote:
> The only question is: does anyone need such a card? Given the ready
> supply of Apple 2 GS computers?

Will the 65816 run at 16mhz in 6502 mode?

1)cheap
2)a 1 bit speaker on card :-)
3)?dma? ?IRQ?

Is the "mailbox" the best approach? Seems slow. All I/O through the card through one register?

Rich

BLuRry

unread,

Sep 26, 2012, 1:45:34 PM9/26/12

to

Well, your choices are to use the mailbox or to contend with bus timing issues.
There could be another way to get around the mailbox limitation or at least expand beyond it a little:

Using the MMU's default behavior, writes to the Cx00 space go to the card and also select the C8-CF banks to also go to the card. If you take advantage of this on the card, you could implement card-side bank-switching so that these memory reads/writes go to selected pages on the card ram. This would let you double the card as a sort of ram expander as well. The first three I/O ports can be used to support "slinky ram" access giving the card much more utility, leaving the other I/O ports to control ram bankswitching.

There would need to be a buffer though -- the motherboard system bus and the card's internal bus are not in sync, and any DMA to the card ram would need to halt the card's bus safely to allow the access. Also, it might be useful to have an option to mark certain accesses to memory to trigger an NMI on the 65c16 side.

Being in slot 7, wouldn't the card also have access to the video bus data? If so, it could be very useful to provide the floating bus value to the 65c16.

All of these are big words, but I know for a fact if I tried any of this I would let a lot of magic smoke out in the process of attempting. :-)

-B