That's like saying "you can't calculate x on a cell phone because it's slower than on a desktop". Yes, there's an exponential slowdown, but that doesn't mean impossible.
The type of slowdown you would get just from moving to a system with a weaker processor (like a cell phone) should be approximately linear. You might get worse-than-linear for certain types of RAM issues (more swapping, etc), but this wouldn't be the same as saying there's an exponential slowdown, and it's not a hard fact about the simulation algorithms themselves.
Anyway, the point isn't that classical computers are just less powerful, the point is that they're so fundamentally different that calculating quantum phenomena can become actually impossible (as in, process will not finish before the heat death of the universe) relatively quickly as the size of the problem increases.
We don't know that, because not all algorithms have been invented. Furthermore, P=NP is an open problem.
Yes, it's not impossible to break 4092bit RSA but it is impossible for all practical purposes [AFAIK]. While it is technically possible to run Crysis on a Commodore it is useless when you have to wait a year to render a single frame.