The ability to simulate the effects of a specific configuration, i.e. SFR content is, I believe, quite critical to a specific MCU implementation. I, for one, like the way in which the Maxim/Dallas 89C4x0's ise tje dual data pointers, with their optional auto-increment/decrement feature.

These MCU's, after all, are just hardware, and if the hardware works differently, a simulator has to reflect the differences accurately. It has to "know" the inner workings of the serial ports, among other features, so that it reflects the interrupt latency correctly. It should, though this one apparently doesn't at the moment, provide accurate oscillator and system clock cycle count, and, ideally, would do so in the context of the various addressing modes in which those DS89C4x0's can operate.

I'm not so sure that those different features are impossible to simulate. After all, the assembler deals with the different features in order to redefine the table of valid SFR's. The same set of SFR definitions can provide a framework for the simulator. Naturally, as the number of available configurations in each MCU increases, the size of that table increases as well.

RE