I haven't seen a proper simulator for Windows, ever.

Everything's been lobotomized in favor of making it "clickable."

Back when folks used HILO, under Unix, AIX, HPUX, or whatever, it did all that stuff, including precise system timing simulation of the processor executing its code set. It even allowed you to vary system timing as the simulated system warmed up, and it allowed for timing variations due to increased or decreased supply voltage and system noise. It could even simulate the ground bounce and other system-wide effects on both supply conditions and signal quality.

Simulation, if properly done, will help you find otherwise unpredictable events caused by peculiarities of timing, such as unanticipated variation in interrupt response time, since you can cause things to happen repeatedly, and at predetermined intervals, that might otherwise occur so infrequently in hardware that you simply don't see them until after the product is shipped.

If a failure doesn't occur until your product is on Mars, it's pretty inconvenient to fix it. Incremental variation or monte-carlo timing simulation will point up sensitivities to such events as might only occur once in 1 billion event cycles.

Even the old DOS-based PSpice allowed various sorts of system-wide effects such as supply noise and thermal effects that affect both digital and analog circuitry.

Simulations thorough enough to find such things are lengthy, but informative.

RE