Re: Modern A2 successor

[Matt Jenkins <mdj@bilby.cs.rmit.edu.au>]

Arkain <rwking@no.usa.spam.net> wrote:
> Not a bad idea. I could use 2 pins to distinguish the running mode:
>     00: Apple2
>     01: GS
>     10: Native
>     11: undefined (room for a 65864)

The 65832 was designed to be a pin compatable drop in replacement for the
65816, and it used the reserved 816 opcode WDM (after the creator) as a
prefix code for 32 bit operations. I think the intention was to provide
a separate MMU for more advanced implementations, but of course that never
actually happened.

Of course, if you were going to redesign it you could give the chip a
proper 24 bit address bus and 32 bit data bus rather than the 8 bit data
bus reused as a bank selector like the 816 which would provide much better
native mode performance.

The beauty of doing this with current day stuff is that you could use 
SDRAM which allows you to have a very simple syncronous RAM interface,
simplifying the design.

The main question is what sort of interface slots you'd have other than
some Apple Bus slots for backwards compatability.

Perhaps a 32 bit version of Apple Bus, which would be very simple to
design.

> and use those to control the memory manager so that the system doesn't have
> to deal with the old-style mangled memory mappings unless it's running in a
> non-native mode.

> 100Mhz should be plenty considering the incredibly small size of A2
> programs. I'll definitely give this idea full consideration. Thanks.

Hell yeah, and you could even run the RAM interface that fast.

Cheers,

Matt

-- 
              Matt(hew) Jenkins. Computer Science (RMIT Bundoora)
                           email: mdj@cs.rmit.edu.au
                            -- Site of the Month --
                            http://www.gnustep.org/