Consider the following scenario. There is a grocery checkout line. The checkout is done with a laser scanner and the results are posted on an electronic board that shows the name, quantity, and cost of the item. But more important is where the items are stored before the checkout. They are kept in a grocery shopping cart.
Let's say that all of the carts are the same size. But not all of the carts are filled to capacity. Let's say that the cart can store 64 items. But some carts only have ten items, some have two, and some have sixty. Each cart has to be handled the same way. Each item has to be processed the same way.
The containers in the cart are part of the process, but handling the cart itself is part of the process, such as identifying the cart, moving it forward, and also replacing the items back into the cart or storing them in a different location, like a grocery bag. So each cart is handled. It is more efficient to have each cart filled to capacity or near capacity, instead of having the items fill only part of the cart, so multiple carts would not have to be used to do the function of one.
This scenario, this analogy to the CPU, explains how efficiencies can be achieved if the right cart is available, a 32 object (bit) cart or a 64 object (bit) cart.