I suspect people who believe in the "life is a simulation" have little experience in running molecular dynamics simulations. Even simulating a single protein for a few microseconds takes a Linux cluster hours. The size of a computer that could simulate every protein on earth (let alone every other type of molecule) would be literally astronomical (as in larger than a planet; and no, plausible improvements in technology really wouldn't shrink that by much). Simulations of things like cities work because all the details of molecules and cells aren't simulated. But we can measure all these things, so either the simulation we are in is really that detailed, or, more plausibly, we aren't in a simulation.