Well, you could just slow down the execution rate of the whole simulation.
But if you limit the speed at which events propagate, my feeling is the rate of events occurring will be lower overall, since one event triggers another and each event will trigger fewer secondary events per second if it propagates less distance per second.
You could now also have islands of stability so that a cataclysm on one end of the simulation will take a long time to spread to the rest of it.
IE, Andromeda can explode and we won't even know for a long time. In that time, we will continue doing interesting / entertaining things, or continue calculating the answer to life, the universe, and everything.
But you will get fewer per step, which is my point. If you get fewer, then you'll need to run the simulation for longer, thus negating any savings.
But you're right in that you'd get different effects, because effects like gravity depend on the distance between objects, which is invariant on the speed of light, so a "human" in a universe where c is halved but everything else is the same wouldn't just be larger, but be completely different.
But if you're changing the behavior of the simulation then it's not even an optimization anymore. You just changed the simulation. It may as well be something entirely different.
But if you limit the speed at which events propagate, my feeling is the rate of events occurring will be lower overall, since one event triggers another and each event will trigger fewer secondary events per second if it propagates less distance per second.
You could now also have islands of stability so that a cataclysm on one end of the simulation will take a long time to spread to the rest of it.
IE, Andromeda can explode and we won't even know for a long time. In that time, we will continue doing interesting / entertaining things, or continue calculating the answer to life, the universe, and everything.