Re: Apple II Simulator

[Mark McDougall <msmcdoug@no.spam.iinet>]

Bryan Parkoff wrote:

>     I have provided my information to the newsgroups -- "Cycle by Cycle 
> Problem?", but I have thought some ideas in my own.  
	(snip snip)

I think you're getting confused between 'cycle accuracy' and real-time 
accuracy (ie. speed of the simulation/emulation).

Generally when you're concerned with cycle accuracy, you're concerned with 
*relative* timings of events within the simulation. For example, you may 
want your cpu core to run different instructions with the exact same timing 
as a real 6502, and have the *exact* "time" elapse with respect to other 
events in the simulation for each instruction.

On the other hand, when concerned with real-time accuracy you're more 
worried about the average speed of the simulation as compared to the real 
thing. So, for example, games would run at the exact same speed as that 
running on a real Apple II.

These are two different goals, one not neccessarily required for the other, 
and have different approaches.

I'm not really sure what you're trying to achieve, and I'm not really sure 
you can actually achieve what you aim to the way I read it (of course I 
could be mis-interpreting you altogether).

Taking the approach of 'simulating' in timeslices (you suggest nano-seconds) 
is similar to circuit simulation and that used for simulation of FPGA 
designs. Generally this approach is very computationally expensive and is 
restricted to sub-components of systems, not whole systems themselves. The 
simulations are based on gate level and/or physical models of components and 
thier interconnections. And in some cases it can take several *hours* to 
simulate a few micro-seconds of design!

Emulators such as MESS don't concern themselves with such accuracy because, 
for most purposes, it is not required. It can afford to allow the CPU to 
execute even a handful of instructions before worrying about the rest of the 
system because there's nothing that really cares too much about relative 
timings within the system. And sure, there's always some exception for some 
tricky piece of software but more often than not - it's by accident rather 
than design!

If you're really interested in cycle-by-cycle simulation of the Apple, I'd 
suggest you look at implementation of the original circuit within an FPGA. 
You can generate all the right clocks on the board and have each part of the 
circuit run exactly as the original circuit did - and in real time too!

Regards,

--
|              Mark McDougall                | "Electrical Engineers do it
|  <http://members.iinet.net.au/~msmcdoug>   |   with less resistance!"