Does unmarked on-street parking accommodate more cars-because smaller cars take less space? Or is unmarked spacing actually less efficient, because of the mismatch effect of very small cars? Should we charge shorter cars less for parking based on their space saving contribution? With many factors in effect, the judgement is not easy. For the first time this study uses computer simulation to assign queues of randomly generated vehicles to marked and unmarked spaces using a parallel world method, to directly test which method is more efficient and can accommodate more cars. Simulation results show that unmarked spacing is more efficient only when the curb is shorter, or not close to any integer times of the optimal length of one marked space, and when drivers are all considerate when choosing the parking location. Under other situations, marked spacing accommodates more cars. This simulation study also found that vehicle downsizing only significantly improves parking efficiency if the vehicle is downsized to two-seaters that can vertically park. In other cases, vehicle downsizing only helps the owners of the downsized vehicles themselves, and making it more difficult for all other people to find parking. To better benefit from vehicle downsizing, a new type of "block-based" spacing is proposed, which achieves some of the flexibility of unmarked spacing and keeping the mismatch-effect relatively low.