Re: FPGApple + ARM?

[bieling@terra.es (Jorge Chamorro Bieling)]

Michael J. Mahon <mjmahon@aol.com> wrote:

> I guess the idea of *dynamically* changing the instruction set
> still seems a little wierd.  I was involved in the design of the
> Burroughs 1700/1800/1900 series, which had dynamic microcode and
> multiple (application language dependent) instruction sets, but
> it wasn't a great idea then, and it still isn't.  

That's because you are thinking of the 6502 as the main processor, but,
what this cpu is is a risc with fixed, not dinamically changing opcodes
in it.

Although, you can play around with the 6502 interpreter code, and do
those things if you wish for any reason, or don't do them if there no
reason to.

When you turn the power on, the arm boots and begins to run an arm
application that happens to be a 6502 emulator that is configurable.
There's always this safe outer shell with a fix, known instruction set,
just in case something goes wrong with the 6502 code that is being
interpreted.

> You need good compilers for each variation that you design, 
> and that is not an easy problem.

It's not. But it's powerful. Better think of it as a configurable
coprocessor if you want to.

> And the performance benefit is negligible, since its all being done in
> software anyway.

Not necesarily true as these opcodes may run in native arm code and they
certainly would execute faster than interpreted 6502 code.
-- 
Jorge Chamorro Bieling