I hold a very similar view, I usually express it like so:

If you were to develop a simulation of a universe you would have to make some concession to be able to run such a large simulation:

• You would have to limit causality, since the communication from one part of the cluster to the next would not be instantaneous you would need to limit the speed at which those communications can happen, that way you guarantee that one part of your cluster can’t interfere with another, think of it like a loading screen.
• Speaking of loading screens, you could make the vast majority of the thing empty, that would limit stuff going over from one part to another.
• You could gain lots of performance by only simulating the micro stuff when required, so an electron could be a wave of possibilities until they need to be somewhere, think of it in the same manner as current games don’t draw what’s not on screen.