... though it is running it in parallel with the target code.

The old ICE's would produce a trace in real time, i.e. the same trace that a step-by-step monitor program would put on the screen is acquired and subsequently displayed, all the while running the target's code in real time. That was an in-circuit-emulator. Clock rates were lower then, but, still, how they managed to do everything that they did was a mystery to me, in view of the fact that they had only that one tentacle with its 40 twisted pairs running between the "emulator" box and the target code.

Functionally, it did everything the monitor does today, only it did it in real time, without breakpoints at the target. That suggests that it ran the target code and captured the result between enpoints of the run, then sifted through the acquired signal history and performed the trace locally before displaying it. I have no idea at all how they managed this, but it was pretty amazing for the time. There were some things that this supposition doesn't explain, though.

RE